Chetana Vara

Applied AI Engineer with 4+ years of experience across Python development, data engineering, machine learning, and backend APIs, including 1+ year building RAG applications and LLM-powered agents. Built and now maintain a production-oriented RAG platform with FastAPI, LangChain, and Pinecone, plus ML pipelines that improved model F1 score from 0.73 to 0.81 and cut API latency from 2.8s to 1.7s. Skilled in document ingestion, embedding/retrieval design, LLM evaluation, structured output validation, and CRM/workflow automation.

AI Engineer - Cognivaton - Remote

(2026-07)

• Maintain a production-oriented RAG platform (Python, LangChain, FastAPI, Pinecone) for client-facing support and lead qualification.
• Extend async FastAPI endpoints, enforcing structured LLM outputs via Pydantic for consistent JSON contracts.
• Hardened reliability with retry/backoff logic, structured logging, and fallback handling for rate limits and malformed outputs.
• Benchmarked OpenAI vs. Claude models on grounding, latency, and consistency to guide model selection.
• Extend HubSpot and Slack integrations (REST APIs, webhooks, n8n) automating lead sync and sales notifications.

AI Engineer Intern - Cognivaton - Remote

(2025-06 - 2026-05)

• Built a RAG-based support and lead-qualification assistant end-to-end using Python, LangChain, Pinecone, OpenAI, and Claude APIs.
• Built a document ingestion pipeline (extraction, chunking, embedding, indexing) with metadata-aware retrieval across PDF, DOCX, and web content.
• Developed FastAPI endpoints and an n8n proof-of-concept extracting lead data, updating HubSpot, and sending Slack alerts.
• Built a 60-question evaluation set and batch-testing workflow tracking citation coverage and failure modes.

Software Engineer – Data & Machine Learning - E Soft US Technologies - India

(2022-02 - 2023-12)

Promoted from Junior Engineer.

• Built Python/SQL pipelines consolidating data across 3 client accounts, processing 50K–100K records/cycle.
• Automated data cleaning and report generation with Pandas/NumPy, cutting manual prep time by ~6 hours/week.
• Built K-Means segmentation models and raised repeat-purchase prediction F1 score from 0.73 to 0.81 via feature engineering.
• Optimized Flask/MySQL reporting APIs, reducing response time from 2.8s to 1.7s.
• Containerized the Flask application with Docker and supported deployment to AWS EC2, with Amazon S3 used for source files and generated reports.

Junior Software Engineer – Python & Data Analytics - E Soft US Technologies - India

(2021-01 - 2022-01)

• Developed Python scripts to clean, merge, and validate multi-source sales and customer data (CSV, Excel, MySQL).
• Ran exploratory data analysis (Pandas, Matplotlib, Jupyter) to surface sales trends and recurring data-quality issues, supporting downstream segmentation models.

Master of Science - Computer Science - Auburn University at Montgomery (2024-01 - 2026-05)

Bachelor of Technology - Computer Science - Ace Engineering College (2017-01 - 2021-01)