Taught & built by Silicon Valley practitioners · Deployed AI at scale

Your job is safe
today.
12 months is
another story.

CoreSmart is built by Agama Solutions — a US IT staffing firm with 20 years of hiring intelligence. They know exactly which developers get kept when teams consolidate around AI. This programme exists because of that gap.

Book a call

Zero card required  ·  No obligation  ·  Pay only if you continue

17
Weeks
16
Named projects
40
Seats · C1
₹0
Week 1
VB
Vinay Bamil
Lead Instructor · Ex-Google GenAI Coach · PhD
✓ Verified
Founded by Agama Solutions
US IT staffing 20 years · hiring intelligence
Cohort 1 starts July 18, 2026
Instructors Ex-Google · Ex-Twitch · Siemens
Seats remaining 40 max · Cohort 1
What industry leaders say

Senior practitioners.
Not students. Not influencers.

These are people who have built and shipped AI systems at scale. Their endorsement is a professional judgement on the curriculum.

★★★★★

"Most AI courses produce people who can talk about AI. CoreSmart produces people who can actually ship. The curriculum covers the same production-grade stack used in real enterprise systems, and every week delivers a usable artifact rather than just a certificate."

AM
Ashish Mago
Co-Founder & CTO, Compliance Kart
Ex CTO, LMCE · Ex Director Technology, Publicis Sapient
↗ LinkedIn
★★★★★

"Teams already have AI tools, but delivery velocity remains unchanged. CoreSmart's dual-track model addresses the real gap by helping Business Analysts specify agent behaviour and Developers govern AI systems at scale."

JM
Jignesh Modi
CxO, Global Hospitals & Healthcare Technology
Digital Transformation & AI Leader
↗ LinkedIn
Why CoreSmart exists

Built by the people who
screen your resume.

Agama Solutions has placed developers in US tech firms for 20 years. CoreSmart was created because they kept seeing the same gap.

2005
Agama Solutions founded
US IT staffing firm begins placing Indian developers in American tech firms. First-hand view of what gets developers hired.
2023
The AI hiring shift begins
Hiring managers start asking for AI engineering in JDs. Most candidates cannot answer the questions. The gap becomes visible.
2025
The consolidation accelerates
42% of code at Indian tech firms is AI-generated. 96% of developers distrust its correctness. The engineer who can validate it becomes hard to replace.
2026
CoreSmart launches
A curriculum built from 20 years of hiring data — not from what educators think the market wants. What hiring managers are testing for right now.
01
The real question
When your company cuts 30% of engineering, who stays?
Not the developer who uses Copilot to write faster. The engineer who can evaluate AI output, govern where it's deployed, and build the systems the team runs on. That person is not replaceable by a tool.
02
What we guarantee
Job readiness. Not a job offer.
CoreSmart does not promise placement. We guarantee you graduate with a deployed system, 16 named GitHub projects, and the specific capability the 2026 hiring market is testing for. The interview is yours to win. The preparation is ours to deliver.
03
Why Agama matters
Demand-pulled curriculum. Not supply-pushed.
Most programmes are built by educators reading the market from outside. CoreSmart's curriculum is pulled directly from what Agama's US clients interview for in 2026 — not what was relevant in 2023. What is being tested for right now.
Before you pay a rupee

Experience the depth first.
Decide after.

Week 1 is free, live, and identical to every paid week. Not a highlights reel. Not a sampler.

Free Week · Zero card · No obligation

One week to decide
if this is different.

5 days of async video content, plus weekend live classes. Same material, same instructors, same cohort as the paid programme. If it's not different from every other course you've taken, walk away. No invoice. No follow-up.

5 days of async video content — same depth as every paid week
Weekend live classes — see how the instructors teach engineering decisions, not slides
The eval framework — how CoreSmart approaches AI reliability from Day 1, not Week 7
Full cohort access — meet the 40 developers you'd build alongside for 17 weeks
₹0
Week 1 · No card required
Zero card · Zero obligation
The instructors

Verify every credential.
LinkedIn. Two minutes.

CoreSmart is a new firm. The instructors are not. Every claim is public and verifiable.

Lead Instructor
Vinay Bamil
Vinay Bamil
Ex-GenAI Coach, Google · PhD AI/ML

Former GenAI Coach at Google with a PhD in AI/ML. Vinay reviews every Cohort 1 capstone personally — evaluating your architecture decisions, engineering tradeoffs, and deployed system. A structural commitment possible only at founding cohort scale.

PhD
AI / ML
Google
GenAI Coach
40
Reviews C1
↗ Verify on LinkedIn
Randeep S. Bhatia
750M
Randeep S. Bhatia
CTO, Splash · Ex-Twitch · Ex-Audible · Ex-EA

Built systems that power 750 million streams. Patent inventor. AAAI published researcher. 100+ conference talks. 15+ years at Twitch, Audible, EA, and Splash.

750M+
Streams
AAAI
Published
100+
Talks
↗ LinkedIn · randeepbhatia.com
Sanjay Lalwani
Sanjay Lalwani
Data Scientist, Siemens · Ex-Infosys

Production ML at Siemens with deep roots in enterprise AI systems at Infosys. Brings the practitioner view — what fails in production and what a large enterprise hiring manager tests for in a real interview.

Siemens
Current
Infosys
Prior
↗ Verify on LinkedIn
What you will build

16 named artefacts.
All on your GitHub.

Not tutorial clones. Named projects with your engineering decisions baked in. A hiring manager opens your GitHub and sees the gap between you and every other candidate.

Week 1
ReleaseBot™
Semantic PR reviewer integrated with GitHub Actions CI. Deployed. Running.
Week 2
EvalKit™
Evaluation pipeline with groundedness scoring and hallucination detection. From Week 2, not Week 7.
Week 3
PromptVault™
Version-controlled prompt registry with diff tracking, rollback, and regression detection.
Week 4
ModelBench™
Benchmark GPT-4o, Claude, and Llama 3 on accuracy, latency, cost, and data residency.
Week 6
CopilotArch™
Working prototype of a GitHub Copilot-style architecture. Five layers. Your decisions.
Week 7
BreakRAG™
Adversarial evaluation harness with LLM-as-judge across 5 providers. Runs as CI.
Week 8
SpecialistTuner™
Fine-tune Llama 3 with PEFT/LoRA. Benchmark vs OpenAI fine-tuning. Write the decision memo.
Week 9
ReliabilityKit™
Hallucination debugging decision tree. The real answer to "What do you do when your LLM hallucinates?"
Week 10
GuardianAI™
PII detection pipeline. Scrubs Aadhaar, PAN, banking data before any LLM call.
Week 11
AgentMesh™
A2A-compatible agent on the live cohort network. Your orchestrator delegates to another student's agent in real time.
Week 12
CostGuard™
Cost profiling and model routing. Identify which 20% of calls consume 60–70% of tokens.
Week 13
IncidentOps™
AI incident response — detection, root cause, rollback, post-mortem. DPDP Act compliant.
Week 14
ObservabilityStack™
OpenTelemetry spans, quality dashboard, cost tracking, anomaly alerts. Minimum viable production observability.
Week 15
MultiModalPipe™
PDF, image, and table processing with modality-specific routing and failure handling.
Weeks 16–17
AgentForge™
The capstone. Five agents, multimodal, PII, MCP, A2A, observability, cost dashboard. Deployed. Live link.
Bonus
PortfolioAgent™
An A2A agent that answers architecture questions about your capstone. Share the link before the interview.

All 16 projects are yours to keep — GitHub repos, your engineering decisions, your name on every commit.

17 weeks · The curriculum

Every week builds
on the last.

5 days async video per week, weekend live classes. From LLM foundations to a deployed, governed, cost-optimised agentic system.

Evaluation instinct starts in Week 2 — not Week 7. Every module from Week 2 onward asks: "How do you know this is working?" Four 90-minute system design case studies run across Weeks 1, 6, 11, and 14 — each produces a one-page architecture diagram.
Phase 1 — Foundations, RAG & Evaluation (Weeks 1–8)
Week 01Modern AI Product Anatomy, LLM Internals & RLHF ConceptsFoundation▼

Opens with live code in hour one — streaming API call, structured output, basic tool call. Topics: Model Layer Architecture, Retrieval + Tool + Memory Layers, Latency-Cost-Quality Triangle, Constitutional AI Concepts, RLHF & Alignment (30-min async reading).

Case Study: How Would You Design ChatGPT? — load balancing, context management, streaming infra, cost attribution, multi-tenant architecture. Produces a one-page architecture diagram.

Project: ReleaseBot — Release Intelligence Assistant. Converts raw update notes into structured changelog + announcement via streaming FastAPI.
Week 02LLM Mechanics, Model Choice & Evaluation PrimerFoundation + Eval▼

Eval primer introduced here — every project from now asks: "How do I know this is working?" Topics: Transformer Behaviour in Practice, Context Windows, Prompting vs RAG vs Fine-Tuning vs Agents, Model Selection Framework, Golden Datasets + Thresholds, Frontier vs Open-Source Decision Criteria, SLM vs LLM.

Project: IntentIQ — Issue Classification Engine. Categorises messages with typed confidence scores. Benchmarks two providers + one open-source model. First golden dataset and accuracy measurement.
Week 03Prompt Engineering, Structured Outputs, Tool Calling & StreamingCore Skill▼

3 progressive async sessions. Day 1 — prompting + structured outputs. Day 2 — tool calling. Day 3 — streaming + retry logic. Topics: System Prompt Design, Few-Shot Patterns, Pydantic v2 Schemas, Output Enforcement Progression (Prompt → JSON Mode → Pydantic → Tool Calling → Structured API), SSE Streaming, Retry & Validation Logic.

Project: TicketStream — Structured Intake Bot with Streaming. Converts messy messages into validated Pydantic objects via SSE streaming.
Week 04Multimodal Ingestion, Embeddings & Vector DatabasesRAG▼

Topics: Embedding Models, Chunking Strategies, Metadata Design, Hybrid Retrieval, GPT-4o Vision + Claude Vision, PDF + Image + Table Ingestion, CLIP / OpenCLIP Embeddings, Nomic Embed (Multimodal), Index Maintenance.

Project: KnowledgeVault — Multimodal Runbook Search. Semantic search over docs, PDFs with figures, tables, and screenshots. Uses CLIP for visual content alongside text embeddings.
Week 05RAG Foundations & Grounded Answers with CitationsRAG▼

Topics: RAG Pipeline Architecture, Citation Generation, Fallback Behaviour, Answer Confidence, Hallucination Controls, Conversation State.

Project: CitationRAG — Engineering Knowledge Assistant. Answers with citations, handles "I don't know" gracefully. Groundedness measured on the golden dataset from Week 2.
Week 06Advanced RAG & Context EngineeringContext Eng.▼

Topics: Query Transformation (HyDE), Step-Back Prompting, Reranking, Multi-Query Retrieval, Context Compression, Before/After Benchmarking.

Case Study: How Would You Design GitHub Copilot? — code embedding, retrieval pipeline, latency constraints, large-repo context management, feedback loops.

Project: RAGOptimizer — Retrieval Quality Lab. Improve Week 5 measurably — query transformation, reranking, compression. Document before/after on latency, recall, groundedness.
Week 07Evaluation Science, Adversarial Testing & ExperimentationEval Science▼

The mature eval harness — eval instinct already established in Week 2. Topics: Inter-Rater Reliability, Evaluator Bias in LLM Judges, Prompt Sensitivity, Statistical Significance in A/B Tests, Dataset Leakage Prevention, LLM-as-Judge Pipeline, Multi-Model Scoring (5 Providers), User Feedback → Eval Loop.

Project: BreakRAG™ — Adversarial Eval & Red-Team Harness. Full evaluation pipeline with adversarial queries, multi-model scoring across 5 providers, regression CI, and user feedback wiring.
Week 08Fine-Tuning, Open-Source LLMs, LoRA Mechanics & SLM SpecialisationOpen Source▼

Includes a 30-min LoRA mechanics explainer: low-rank decomposition, rank 8 vs rank 64 tradeoffs — genuine mechanical intuition before the code. Topics: When Fine-Tuning Wins, PEFT/LoRA on Llama 3/Mistral, Hugging Face Training Pipeline, Ollama Local Inference, OpenAI Fine-Tuning API (GPT-4o-mini), SLM Routing, Proprietary vs Open-Source Cost/Control/Privacy.

Project: SpecialistTuner — Dual Fine-Tuning Lab + Decision Memo. Fine-tune via PEFT/LoRA on Llama 3 AND OpenAI fine-tuning API. Benchmark both on accuracy, latency, cost, data privacy. Decision memo: when does each approach win?
Phase 2 — Agentic AI Engineering (Weeks 9–13)
Week 09Agentic AI Foundations & Agent State ManagementAgentic▼

Build the agent loop from first principles before touching any framework. The "when NOT to use agents" case study is weighted equally to the build. Topics: Workflows vs Agents (with explicit Reject exercise), Tool Loops & Stop Conditions, Agent Design Patterns (ReAct, Reflection), Short-Term / Workflow / Persistent State, State Diagrams as Deliverables.

Project: OpsAssist — Tool-Using Ops Agent + State Diagram. Single agent with defined tool schemas, stop conditions, retry logic. Full state diagram showing three state layers is a required deliverable.
Week 10Multi-Agent Orchestration, Long-Term Memory & When NOT to Over-EngineerAgentic▼

Explicit "reject multi-agent" case study — students must argue against multi-agent for a specific scenario. Topics: Routing & Handoffs, Manager-Worker Patterns, Graph Orchestration (LangGraph), OpenTelemetry Tracing, In-Context Memory, External Memory (Vector Store), Episodic Memory, Procedural Memory, Memory Lifecycle Management.

Project: TriageFlow™ — Incident Multi-Agent Workflow + Memory. Triage → knowledge → action agents with full trace replay, human approval gate, and four-layer memory (Redis session, pgvector long-term, episodic summaries, procedural preferences).
Week 11Agent Interoperability — MCP, A2A & AGENTS.mdMCP + A2A▼

Topics: MCP Architecture & Tool Design, Publishing MCP Servers, A2A Protocol (Agent Cards), Client-Remote Architecture, Task Lifecycle & SSE Streaming, JWT/OIDC Security, AGENTS.md Spec Files.

Case Study: Production Multi-Agent System Design — orchestration at scale, agent isolation, handoff protocols, trace architecture, failure recovery, cost visibility.

Project: AgentMesh™ — A2A Remote Specialist + Shared MCP Server. Publish a signed Agent Card, expose A2A-compatible service with SSE + JWT, AND publish one MCP server the cohort consumes.
Week 12Guardrails, PII Detection, Responsible AI & HITLSecurity + Ethics▼

Topics: Prompt Injection Defense, Tool Misuse Prevention, Approval Checkpoints + Audit Logging, PII Detection (Microsoft Presidio), NER-Based Entity Scrubbing, Output PII Filtering, PII in RAG Pipelines, Responsible AI Framework, Bias & Fairness Awareness, HITL Design Patterns, GDPR/CCPA in AI Systems.

Project: GuardianAI™ — Adversarial Lab + PII Shield + Responsible AI Checklist. Students attack and defend in pairs. Add Presidio PII scrubbing on inputs + outputs. Complete formal responsible AI checklist with HITL design, bias assessment, GDPR notes.
Week 13Developer Tooling Agents, Streaming Operator UI & UX Feedback LoopsDev Tools▼

Topics: Repo Intelligence Agents, CI/CD Context + PR Assistance, Streaming Chat UI (SSE + React), Citation Rendering, Tool Trace Display, Approval Action UI, Thumbs Up/Down + Correction Flows, Feedback Storage → Eval Harness, Online vs Offline Eval Loop.

Project: WorkbenchAI™ — Streaming Developer UI + Feedback Loop. Portfolio-quality UI: live citations, tool traces, approval actions. Wires the full user feedback loop — thumbs/corrections → feedback database → BreakRAG™ eval harness.
Phase 3 — Production, Reliability & FinOps (Weeks 14–16)
Week 14Deployment, Observability, Versioning & Async PipelinesLLMOps▼

Topics: Docker + CI/CD, Cloud Deployment, OpenTelemetry + Tracing, Monitoring + Alerting, Model + Prompt Versioning, Dataset Versioning, Async Background Jobs, Batch Processing Pipelines.

Case Study: AI Observability System Design — token-level cost tracking, latency percentiles, model drift detection, eval regression, incident response.

Project: DeployCore — Production Deployment with Background Jobs. Cloud-deployed agent with OpenTelemetry, async ingestion/eval pipelines, and versioning controls.
Week 15AI Application Reliability EngineeringReliability▼

This week gets its own space — not compressed with deployment. Topics: Hallucination Debugging Decision Tree, Prompt Versioning in Git, Model Upgrade Control (A/B, Frozen Test Sets), Pydantic Response Enforcement Patterns, Parsing Fallback Strategies, Retry + Backoff Architecture, When Structured Output Breaks.

Project: ReliabilityKit™ — AI Reliability Engineering Toolkit. Hallucination debugging decision tree, Git-based prompt versioning workflow with regression checks, model upgrade protocol with frozen test sets, Pydantic enforcement patterns, fallback strategies.
Week 16FinOps, Prompt Caching, SLM Routing & Event-Driven Self-HealingFinOps▼

Students measure and optimise their actual costs on their actual workload. Topics: Prompt Caching (Anthropic/OpenAI), Intelligent SLM/LLM Routing, Per-Request Cost Budgets, Cost Dashboards + Measurement, Event-Driven Webhook Triggers, Agentic SRE Patterns, Pause/Resume + Human Approval.

Project: CostGuard™ — FinOps Router + Self-Healing SRE Bot. Cost-aware routing with prompt caching and weekly cost measurement report. Plus an event-driven SRE agent that diagnoses root cause and proposes remediation with human approval.
Phase 4 — Capstone Hardening & Demo Day (Week 17)
Week 17Final Capstone Hardening, Demo Day & Hiring PackCapstone▼

Topics: Production Polish, Full Red-Team Review, Eval + Reliability Report, Architecture Case Study, Responsible AI Checklist, Cost Analysis, Technical Storytelling.

Project: AgentForge™ — Final Demo Day. Live demo, architecture review, eval report, ReliabilityKit™ walkthrough, security checklist audit, cost analysis. Bonus: PortfolioAgent — A2A-compatible, knows your capstone, answers architecture questions, deployed as a shareable link for hiring managers.
The maths · Your numbers

How long does the fee take to recover?

Enter your numbers. We show you the payback period. No salary guarantee — just arithmetic. Verify it yourself.

₹8,333
Monthly delta
₹274
Daily increase
~14.4 mo
Payback period

These are your numbers on your inputs. CoreSmart makes no salary guarantee. The job offer depends on the interview. The capability to earn it is the CoreSmart commitment.

Pricing · Transparent · No hidden fees

Three tiers. Every item listed.
No "contact for pricing."

The free week costs nothing. After that, the decision is yours — with full information.

Best Value
Early Bird
₹85,000
One-time · All taxes included
  • ✓17 weeks — async video + weekend live classes
  • ✓All 16 named projects + AgentForge™ capstone
  • ✓5 days async video + weekend live classes per week
  • ✓Cohort A2A network access
  • ✓Resume bullets with your capstone metrics
  • ✓GitHub repos — all yours to keep
No placement guarantee. Job readiness through a deployed system and verifiable capability.
Standard
₹1,25,000
One-time · All taxes included
  • ✓Everything in Early Bird
  • ✓Group office hours with Vinay — 2× per week
  • ✓Capstone reviewed by Vinay personally
  • ✓PortfolioAgent™ — your pre-interview AI representative
  • ✓Agama Solutions hiring manager brief
  • ✓Priority alumni network access
No placement guarantee. Job readiness through a deployed system and verifiable capability.
Premium
₹1,75,000
One-time · All taxes included
  • ✓Everything in Standard
  • ✓3× 1-on-1 sessions with Vinay (60 min each)
  • ✓Mock technical interview with Randeep Bhatia
  • ✓Week 18 Computer Use Lab (Premium only)
  • ✓Direct intro to Agama Solutions hiring contacts
  • ✓Lifetime cohort alumni access
No placement guarantee. Job readiness through a deployed system and verifiable capability.
Cohort 1 · July 18, 2026

40 developers.
Vinay reviews every capstone personally.
This is a structural commitment.

Cohort 1 is 40 developers. Vinay Bamil reviews every capstone personally — your architecture decisions, your engineering tradeoffs, your deployed system. This will not be true for Cohort 5. The instructor-to-student ratio that makes this possible exists only now. The founding cohort also shapes the curriculum in real time.

40
Max seats
Jul 18
Start date
Honest answers

The objections.
Answered directly.

The fee is high. I've paid for courses before and got nothing.▼
That is the correct objection. This is why Week 1 is free and identical to every paid week — not a sampler. Experience the actual depth, the actual eval framework, the actual live session format. If it's not different from every other course, walk away. No invoice. The fee recovers when your CTC moves — use the ROI calculator with your own numbers.
Is there a placement guarantee?▼
No. CoreSmart does not guarantee a job. We guarantee you graduate with a deployed system, 16 named projects on GitHub, and the engineering capability to demonstrate — not describe — what you built. Any programme that claims placement guarantee in the Indian market is either lying or burying the cost somewhere you can't see it.
My job feels secure right now. Why act now?▼
It probably is. The risk is not today — it's 12 months from now when the engineer who can build, evaluate, and govern AI systems has become the standard expectation at your level. Building that capability takes 17 weeks. Starting now means you're ahead of the consolidation, not behind it. The free week costs nothing and answers the question in 7 days.
CoreSmart is new. Why should I trust you?▼
Verify the instructors. Vinay Bamil — ex-Google GenAI Coach, PhD, LinkedIn-verifiable in 2 minutes. Randeep S. Bhatia — CTO Splash, ex-Twitch, ex-Audible, randeepbhatia.com. Sanjay Lalwani — Data Scientist Siemens, LinkedIn-verifiable. Agama Solutions — 20-year US IT staffing firm. Every claim is public. If a single one is wrong, the free week costs you nothing.
I already use AI tools every day. What more do I need?▼
96% of developers distrust AI-generated code correctness (SonarSource 2025). The skill the market is pricing in 2026 is not generation — it's validation. Evaluation pipelines, hallucination debugging, model upgrade control, adversarial testing. Using AI makes you a user. Building and governing AI systems makes you the engineer who is harder to replace.
I have 2 years experience. Is this too advanced?▼
CoreSmart is for developers with 2–6 years who already ship production code. The programme builds the AI engineering layer on top of what you already know. The free readiness assessment tells you honestly if the timing is right.
Why does the live cohort format matter?▼
In Week 11, your orchestrator discovers and delegates to another student's agent in real time in a live cohort network. A self-paced recording cannot replicate that. AgentMesh™ requires 40 developers building together. That artefact — an A2A-compatible agent that worked with a real agent in a real live network — is not reproducible alone.
Free Week · Cohort 1 · July 18, 2026

Start your free week.
No card. No obligation.

Week 1 — 5 days async video, weekend live classes. Same instructors, same cohort, same depth as the paid programme. Walk away if it is not the right fit.

Book a 20-minute call