Résumé ↗
Agentic AI Engineer · building context-aware GenAI systems

Venkatesh G

Scroll ToExplore

03 // Some things I've built

Projects in orbit

▸ CLICK A CARD TO INSPECT

Explore the archive ↗

Initializing Launch Sequence

0%

Venkatesh G — Agentic AI Engineer | GenAI Systems & Context Engineering

Agentic AI Engineer · building context-aware GenAI systems

Agentic AI Engineer building production GenAI systems across multi-agent orchestration, context engineering, RAG, MCP, evaluation, and observability.

About

Most AI prototypes look great until they hit real edge cases and enterprise scale. I focus on building reliable agentic workflows and retrieval architectures that can plan multi-step tasks, query tools safely, and recover cleanly when inputs are messy.

My day-to-day work centers on LangGraph, Model Context Protocol (MCP), and GraphRAG pipelines. I care about what happens under the hood: structured schema enforcement, deterministic tool routing, and token-level evaluation to ensure systems behave consistently.

I built AgentOps Harness around that exact mindset. It gives autonomous coding agents a verifiable control plane by grounding every action in diff attribution, test runs, and deterministic execution logs instead of unchecked model output.

Based in Houston, TX, United States.

Experience

AI Engineer — Infosys

May 2025 — June 2026 · Houston, TX

Built an enterprise multi-agent platform that turned L2 batch-job failures into deterministic, auditable triage workflows.

  • Reduced investigations from hours to minutes for 10+ engineers across Autosys, SQL, file servers, and ServiceNow
  • Built four secured MCP tool servers with Azure Active Directory and enterprise token management
  • Used GraphRAG to retrieve domain SOPs and produce structured plans with explicit dependencies
  • Built a LangGraph graph generator that parallelizes independent steps and dispatches specialized agents
  • Instrumented token-to-decision traces with Phoenix and ITrace for auditability and compliance

AI / NLP Engineer — X Node Inc.

Aug 2024 — May 2025 · Remote

Delivered production GenAI agents, RAG systems, and NLP pipelines for enterprise workflows.

  • Built LangGraph agents using ReAct and plan-and-execute orchestration
  • Improved RAG retrieval with Pinecone, FAISS, Azure OpenAI embeddings, hybrid search, and reranking
  • Engineered PII masking, NER, and entity-linking pipelines for legal-domain privacy requirements
  • Fine-tuned BERT and RoBERTa on Vertex AI for classification and conversational tasks
  • Reduced debugging time about 40% through token-level execution tracing and retrieval diagnostics

Data Engineer (Contract) — Walmart

Jun 2023 — Jul 2024 · Bentonville, AR

Built high-volume batch and streaming data systems for analytics and data-science teams.

  • Maintained PySpark and Scala ETL pipelines processing more than 100GB daily
  • Delivered near-real-time streaming with Spark Streaming and GCP Pub/Sub
  • Reduced storage costs about 25% while improving warehouse query performance
  • Accelerated insight generation about 60% through tailored cross-functional datasets

Data Analyst — Capital Numbers

Jun 2019 — May 2021 · Noida, India

Built recommendation, ingestion, analytics, and model-evaluation workflows in Python and Spark.

  • Built a Python recommendation model with custom feature engineering
  • Designed Spark and Kafka ingestion for large structured and unstructured datasets
  • Created dashboards that increased actionable insights about 25%
  • Automated model evaluation with custom Python tooling

Projects

AgentOps Harness

Agentic systems · Open source — Evidence-backed control plane for coding agents

A LangGraph control plane that scans repositories, plans work, hands off to interchangeable coding agents, attributes diffs, runs tests, scores risk, blocks unsupported claims, and produces a PR-ready report.

Built with: Python, LangGraph, Typer, FastAPI, MCP, SQLite.

NL2SQL Viz

Analytics · AI safety — Natural language to guarded SQL and charts

An end-to-end analytics product spanning schema introspection, Claude-powered SQL generation, deterministic read-only guards, Postgres execution, chart planning, and Vega-Lite rendering.

Built with: Python, FastAPI, Claude, Next.js, Vega-Lite.

Enterprise Agentic Graph RAG

Graph RAG · Evaluation Harness — Verifiable Graph RAG with Calibrated DeepEval Proving Ground

A multi-agent Graph RAG platform over SEC 10-K filings featuring typed fact graphs with chunk provenance, safe Python calculation tools, and a two-tier DeepEval harness benchmarked against FinanceBench (81.3%) and ConvFinQA.

Built with: LangGraph, FastAPI, FastMCP, DeepEval, BGE-Reranker, Qdrant.

Skills

Agentic AI

Production agent systems that plan, coordinate, call tools, preserve state, and return structured, traceable outcomes.

  • LangGraph
  • LangChain
  • Pydantic AI
  • Semantic Kernel
  • Instructor
  • MCP
  • Multi-agent systems
  • ReAct agents
  • Planning agents
  • Tool calling
  • Structured outputs
  • Memory
  • Long-context orchestration

Context Engineering & RAG

Retrieval and context pipelines that assemble the right evidence, instructions, and model context for grounded GenAI behavior.

  • Context engineering
  • GraphRAG
  • Agentic RAG
  • RAG architecture
  • Prompt engineering
  • Context-window management
  • Claude
  • Azure OpenAI
  • Vertex AI
  • Hugging Face

Models & NLP

Applied language-model and NLP workflows for understanding, retrieval, classification, and grounded generation.

  • PyTorch
  • Transformers
  • BERT
  • RoBERTa
  • Classification
  • NER
  • Semantic search
  • Entity linking
  • Summarization
  • Question answering

Languages & Data Systems

Languages, processing systems, and analytics tooling used to build dependable AI data paths and production workflows.

  • Python
  • TypeScript
  • SQL
  • Pandas
  • NumPy
  • PySpark
  • Kafka
  • Airflow
  • Snowflake
  • Matplotlib
  • Seaborn
  • Plotly

Databases, Vectors & Cloud

Storage, vector retrieval, cloud, container, and delivery infrastructure for production GenAI systems.

  • Postgres
  • MongoDB
  • Qdrant
  • Pinecone
  • FAISS
  • Azure AD
  • Azure Kubernetes Service (AKS)
  • GCP BigQuery
  • GCP Pub/Sub
  • AWS
  • Docker
  • Kubernetes
  • Jenkins CI/CD

Observability, Evaluation & Safety

Evidence, evaluation, safety, and developer tooling for operating GenAI systems with measurable quality and controlled risk.

  • Phoenix
  • OpenTelemetry
  • LangSmith
  • RAGAS
  • DeepEval
  • LLMOps
  • Evaluation design
  • Guardrails
  • Human-in-the-loop
  • OWASP AI security
  • Token economics
  • Claude Code
  • Cursor
  • GitHub Copilot
  • Hermes
  • Pi

Contact

Email: venkatesh.gtd1@gmail.com

Phone: +1 660-853-9110

Location: Houston, TX, United States