Skip to content

Vinay Gundela

Generative AI Engineer • Houston, TX • v**************@gmail.com • 415****775 • drivetube.ai/•••••

Professional Summary

Generative AI Engineer with 5+ years of experience and 3+ years building and deploying production GenAI and LLM-based solutions, including conversational AI, agentic workflows, and GenAI-driven data analytics pipelines. Skilled in prompt engineering, retrieval-augmented generation, LangChain/LangGraph orchestration, and production API design using FastAPI and Docker. Experienced turning unstructured and multimodal data into structured, decision-ready outputs and implementing observability and rule-driven validation for deployed systems.

Technical Skills

Programming Languages: Python,JavaScript
Frameworks and Libraries: LangChain,LangGraph,Agentic design patterns,Function,tool calling,Memory management,FastAPI,Flask,Django,gRPC,REST API design,Pandas,NumPy
Databases: SQL,Pinecone,FAISS,pgvector,MongoDB Atlas Vector Search,Weaviate,Redis,MongoDB,GridFS,MySQL,PostgreSQL
Cloud and DevOps: AWS,Azure,Azure OpenAI Service,Docker,Kubernetes,GitHub Actions,CI,CD
Data and Analytics: Model evaluation,ETL pipelines,Anomaly detection
Tools and Methodologies: Agile,Scrum,Code review,Rate limiting and retry patterns
Generative AI & LLMs: OpenAI GPT-4, GPT-4o,Anthropic Claude,Google Gemini,LLaMA,Mistral,Prompt engineering
Retrieval & Indexing: Retrieval-Augmented Generation,Embedding generation,Chunking strategies,Semantic search,Hybrid search
Multimodal & Document Intelligence: Vision-Language Models,OCR,PyMuPDF,Signature matching,Bounding-box validation
Distributed Systems & Queues: Celery,RabbitMQ,AsyncIO,Microservices
Monitoring & Observability: RAGAS-style monitoring,Tracing and logging,Verdict,evidence persistence patterns

Work Experience

LTIMindtree
Houston, TX
Generative AI Engineer
Mar 2025 – Present
IT services engagement building GenAI-driven analytics and verification features for a multi-market document platform serving Turkey, India, and Malaysia, integrating .NET microservices with Python GenAI processors.
Tech Stack: LangGraph, Celery, Redis, Python, MongoDB, OCR, VLMs, YAML, Docker
  • Architected and shipped Generative-AI-driven data analytics and verification features for a polyglot, multi-market document platform integrating seven .NET data-layer microservices with a Python GenAI processor, enabling model-agnostic analytics workflows.
  • Designed and implemented a 4-node LangGraph agentic orchestration flow coordinating ingestion, validation, parallel document processing, and finalization across Celery task queues with Redis-backed chord coordination to enable resilient parallel processing.
  • Implemented a two-stage Vision-Language Model perception pipeline (visual summary then field extraction/classification) to convert unstructured and handwritten financial documents into structured, analytics-ready records using OCR + LLM extraction.
  • Engineered a YAML-driven configuration framework for multi-market analytics rules, enabling new market onboarding through config-only changes and reducing onboarding effort from weeks to days via reusable templates and deployment scripts.
  • Authored Malaysia market data validation and extraction specification for private CDR flow, defining 22 single-document and cross-document rules across hard (Tier 1) and score-weighted (Tier 2) tiers applying structured output prompts and governed rule logic.
  • Tuned signature-matching thresholds and bounding-box overlap heuristics in the decision orchestrator and implemented a two-artifact MongoDB persistence pattern (verdict output and full evidence graph) to improve observability and reduce time-to-root-cause for incidents.
Genpact
Hyderabad, India
Gen AI Engineer
Sep 2020 – Mar 2023
Professional services role building and integrating conversational and generative AI solutions for enterprise and SMB clients, including document-based knowledge systems and agentic workflows.
Tech Stack: OpenAI, Anthropic Claude, Google Gemini, LangChain, Pinecone, FastAPI, Flask, Pandas, Python
  • Built and deployed conversational and generative AI applications using OpenAI GPT-4, Anthropic Claude, and Google Gemini APIs integrated with client systems across enterprise and SMB customers to deliver document Q&A and automation features.
  • Developed Retrieval-Augmented Generation pipelines with LangChain and Pinecone for document-based AI systems used for internal knowledge bases and enterprise Q&A, focusing on chunking strategies and embedding quality.
  • Implemented agentic, tool-using conversational workflows with LangChain Agents to enable multi-step task automation and structured decision-making inside client workflows.
  • Built scalable REST APIs with FastAPI and Flask including rate limiting and retry logic for reliable LLM call management and production-grade error handling.
  • Applied prompt engineering techniques (few-shot, chain-of-thought, structured outputs, function calling) and evaluation strategies to reduce hallucinations and improve response reliability in production integrations.
  • Prepared and cleaned client datasets with Pandas-based ETL pipelines, identified anomalies for downstream modeling, and debugged production issues related to latency, token limits, and retrieval quality.
Nacre Software Services Pvt Ltd
Hyderabad, India
Data Scientist
Sep 2019 – Aug 2020
Data science role focused on conversational AI and data preparation for analytics; supported internal chatbot and modeling initiatives.
Tech Stack: NLTK, Pandas, NumPy, Scikit-learn, Python
  • Built an intent-classification conversational AI chatbot using NLTK and custom preprocessing, achieving ~82% accuracy and deploying the system for internal production use.
  • Cleaned and preprocessed a ~50,000-record customer dataset using Pandas and NumPy to improve data readiness and feature quality for downstream models.
  • Supported senior data scientists with model evaluation and hyperparameter tuning for Random Forest and logistic regression models, contributing to measurable improvements in validation metrics.
  • Developed visualizations and dashboards to communicate model results and insights to non-technical stakeholders, enabling data-driven business decisions.
  • Documented data pipelines and model evaluation procedures to standardize reproducibility and handoff to operations teams.
  • Collaborated in cross-functional reviews to prioritize modeling work aligned with internal business requirements and production constraints.
Zensar Technologies
Hyderabad, India
Junior Software Developer
Jan 2019 – Jun 2019
Entry-level backend development role building REST APIs and data services for internal business applications within an Agile SDLC.
Tech Stack: Flask, Django, MySQL, PostgreSQL, Git
  • Built REST APIs using Flask and Django for internal applications, handling user data, authentication, and core CRUD operations to support business workflows.
  • Worked with MySQL and PostgreSQL for relational data modeling, query optimization, and reporting use cases, improving data access performance.
  • Implemented standard Git-based version control and participated in code reviews to maintain code quality and team standards.
  • Participated in Agile ceremonies including sprint planning, daily stand-ups, and retrospectives, gaining exposure to SDLC best practices.
  • Collaborated with QA and product owners to deliver incremental features and fixes on schedule, supporting deployment activities.
  • Created backend endpoints and data access layers that enabled reporting and analytics features consumed by internal stakeholders.

Projects

Intelligent Document Q&A
Tools Used: LangChain, Pinecone, GPT-4, FastAPI, Redis, PyMuPDF, RAG
  • Built a Retrieval-Augmented Generation pipeline for document Q&A with sub-500ms retrieval and Server-Sent Events streaming to deliver low-latency responses.
  • Implemented multi-tenant isolation and RAGAS-style monitoring for production observability and retrieval-quality tracking.
  • Integrated PyMuPDF for document parsing and used Pinecone for vector indexing to support semantic search at scale.
AskSmart Multi-Model Platform
Tools Used: GPT-4, Anthropic Claude, Google Gemini, FastAPI, Model routing
  • Built a multi-model orchestration platform with complexity-based routing across GPT-4, Claude, and Gemini to optimize cost and latency.
  • Implemented fallback chains, retry logic, and a unified LLM interface in FastAPI, reducing API spend by ~30% through model selection strategies.
Agentic Contract Workflow
Tools Used: LangGraph, Clause extraction, Risk scoring, Fault tolerance
  • Designed a LangGraph-based pipeline for clause extraction, risk scoring, and conditional routing (approve/reject/human-review) with a checkpointer for fault-tolerant processing.
  • Automated conditional decision paths and integrated risk scoring to reduce manual review workload in contract processing flows.

Education

Golden Gate University
MS in Information Technology Management • California, USA • May 2023 – Dec 2024

Certifications

OpenAI API Development
Advanced Prompt Engineering
Full Stack Data Science & AI Program
Python and Django Full Stack Web Developer
NASSCOM Certificate

Powered by Drivetube · Create your own profile at drivetube.ai