╭────────────────────────────────────────────────────────────────╮
│ ● ● ● profile.sh │
├────────────────────────────────────────────────────────────────┤
│ shabeeth2@github:~$ ./whoami │
│ │
│ __ __ ___ _ _ _ __ __ ___ ___ │
│ | \/ |/ _ \| || | /_\ | \/ | __| \ │
│ | |\/| | (_) | __ |/ _ \| |\/| | _|| |) | │
│ |_| |_|\___/|_||_/_/ \_\_| |_|___|___/ │
│ │
│ │
│ ___ _ _ _ ___ ___ ___ _____ _ _ │
│ / __| || | /_\ | _ ) __| __|_ _| || | │
│ \__ \ __ |/ _ \| _ \ _|| _| | | | __ | │
│ |___/_||_/_/ \_\___/___|___| |_| |_||_| │
│ │
│ │
│ AI Engineer / RAG + agents + backend │
│ Chennai, India / TCS │
╰────────────────────────────────────────────────────────────────╯
LANGUAGES Python · Java · SQL
AI LangChain · LangGraph · MCP
RAG Gemini · Vertex AI Vector Search · RAGAS
API / DATA FastAPI · Spring Boot · PostgreSQL
DELIVERY Docker · Jenkins · GitHub Actions · Git
Python / Jupyter Notebook / TypeScript / JavaScript / Java
Python and notebooks lead the public repository list. Languages describe code, not skill proficiency. SQL is listed above as a skill, not a GitHub primary-language percentage.
|
Data investigation with tool-calling agents, read-only SQL and evidence-backed dashboards.
|
Speculative read-only tool execution during streamed generation, with exact-cache reuse and final-script replay.
|
|
A tool for reviewing agent work as HTML artifacts visually, human or AI-generated: element-level annotations, chat, revisions and persistent review sessions.
|
|
ML-Agents · 19.8k starsOpen-source contributor: documentation fixes for the TerminalSteps guide.
|
Apprise · 17.6k starsOpen-source contributor: documentation fixes.
|
txtai · 13k starsOpen-source contributor: workflow input example corrected to match the existing code sample.
|
yargs · 11.5k starsOpen-source contributor: shell-completion install example that appends to shell configuration instead of overwriting it.
|
attrs · 5.9k starsOpen-source contributor: documentation fixes.
|
|
Talk2Data ·
SpecExec ·
Duet
Open-source contributor to
ML-Agents ·
Apprise ·
txtai ·
yargs ·
attrs
About, project details, skills and qualifications
AI Engineer at Tata Consultancy Services, Chennai, India.
I build AI application components for production enterprise RAG and backend integrations. My RAG work covers retrieval, prompts, grounding checks, guardrails and evaluation using LangChain, Gemini and Vertex AI Vector Search. My contribution is the AI application layer, not cloud infrastructure or operations ownership.
I also build Java and Spring Boot modules with Kafka for production data migration and synchronization, and work on regression validation for migration releases. My ticket-assistance work is a prototype, with classification, similar-case retrieval and human review.
Employment: Tata Consultancy Services, Sep 2024 - Present.
- Winner, AI Fridays Innovator Award, TCS (2025).
- On the Spot (Team) Award, TCS, 3 Mar 2026, for a rapid prototype built and presented as a team.
- tcsAI Idea Igniter, TCS, 4 Nov 2025: hackathon idea selected among top entries.
- Delivered GenAI and AI engineering sessions for TCS associates.
A local data investigation workbench with tool-calling agents, schema inspection, validated read-only SQL, evidence-backed charts and dashboards. Built with Python, FastAPI, Next.js and MCP. Uses step and tool-call limits and explicit approval before MCP calls.
https://github.com/shabeeth2/talk2data-agents
A Python runtime that executes eligible read-only tool calls while generation streams, then replays the final script from a fresh namespace and reuses exact cache hits. A local simulated benchmark on independent calls measured 764 ms to 368 ms median latency, a 51.8% reduction. This is not a production performance claim.
https://github.com/shabeeth2/specexec
A tool for reviewing agent work as HTML artifacts visually, human or AI-generated. Review sessions track element-level annotations, chat, revisions and approval, with live updates, layout checks and a self-contained HTML export.
https://github.com/shabeeth2/duet
- Open-source contributor to ML-Agents (19.8k stars): documentation fixes for the TerminalSteps guide.
- Open-source contributor to Apprise (17.6k stars): documentation fixes.
- Open-source contributor to txtai (13k stars): workflow input example corrected to match the existing code sample.
- Open-source contributor to yargs (11.5k stars): shell-completion installation example corrected to append instead of overwriting shell configuration.
- Open-source contributor to attrs (5.9k stars): documentation fixes.
Python, Java, SQL, LangChain, LangGraph, Gemini on Vertex AI, OpenAI API and embeddings, Vertex AI Vector Search, ChromaDB, hybrid retrieval, reranking, RAGAS, FastAPI, Spring Boot, Kafka, Docker, Jenkins, GitHub Actions, Redis, Grafana, Linux and Git.
B.Tech, Artificial Intelligence and Data Science, SNS College of Engineering, Coimbatore. Graduated May 2024. CGPA: 8.4/10.
- Claude Certified Architect - Professional, Anthropic. Issued Aug 2026; valid through Aug 2027.
- Google Cloud Generative AI Leader. Issued Feb 2026.
- Microsoft Certified: Azure AI Fundamentals (AI-900). Issued May 2024.
https://linkedin.com/in/shabeeth
Snapshot: 5 Oct 2026. GitHub language-byte shares are normalized inside each public, non-fork repository with detected code, then averaged with equal repository weight (23 repos). Repositories without detected code are excluded. This chart describes code composition, not skill or proficiency. Other includes HTML, JavaScript and Shell. Bars are static until regenerated.