Harsh Vardhan Tripathi

AI / ML Engineer · Product Engineer

Lucknow, Uttar Pradesh, India

@harsh_vardhan_tripathi

LinkedIn GitHub

0 followers

🎓 BS in Data Science and Applications at Indian Institute of Technology Madras · Graduating 2027

About

AI/ML and product engineer experienced in building and deploying generative AI, LLM, agentic AI, and backend systems. Works across AI architecture, prompt engineering, RAG, model integration, production deployment, and technical leadership.

Experience

Product Engineer

Sentio Mind · Lucknow, India

Sep 2026 – Present

Engineered AI-powered product features across backend services, APIs, databases, AI integrations, and production applications. Built backend workflows for field-employee visit management, GPS/geofence verification, visit dispatch, forms, check-in/check-out, and map-based tracking. Designed, implemented, tested, and deployed production backend and AI systems across client projects, including API integrations, database workflows, system debugging, and reliability improvements. Worked as a technical consultant across client requirements and product development, translating business requirements into scalable software and AI solutions and coordinating implementation with engineering teams.
AI Developer

Banao Technologies · Lucknow, India

Nov 2025 – Aug 2026

Re-engineered the Interview God AI evaluation architecture to improve reliability and consistency of automated interview evaluation through structured processing, validation, and robust evaluation workflows. Designed the architecture for an AI-powered call screening system and an AI-based coding evaluation system. Led a team of AI developers, driving technical execution, architecture decisions, development priorities, and delivery of AI initiatives.
AI Intern

Innovate · Remote, India

Jun 2025 – Jul 2025

Built a fully offline chatbot using Qwen 0.5B, Hugging Face Transformers, and PyTorch, implementing a 4-turn sliding memory window. Achieved 2–5 second CPU inference latency without GPU dependency through optimized model loading and inference. Developed dual CLI and Streamlit interfaces with modular architecture and optimized memory management for resource-constrained environments.
Machine Learning Intern

Edunet Foundation · Remote, India

Apr 2025 – May 2025

Trained ResNet50 using transfer learning for plant disease classification across 38 classes using a curated dataset of 50,000+ images. Built the end-to-end deep learning pipeline including data augmentation, normalization, class-balanced batching, training, and evaluation. Evaluated model performance using accuracy, precision, recall, and F1-score and documented results for technical reporting.

Education

2026

B.Tech in Computer Science and Engineering Verified

Babu Banarasi Das Institute of Technology & Management

Awarded by AKTU

Computer Science and Engineering (CSE) · Nov 2022 – Aug 2026

Computer Science and Engineering (CSE) · Nov 2022 – Aug 2026

Skills

<3 months

C SQL RAG LLMs Bert Keras Python Tf-Idf PyTorch Resnet-50 JavaScript TensorFlow Fine-tuning scikit-learn Deep Learning LLM evaluation Computer Vision Model Evaluation Transfer Learning Prompt Engineering Agentic AI Systems Agentic AI workflows Natural Language Processing Retrieval Augmentation Generation Generative AI (ChatGPT, Copilot, Gemini)

Tools / apps / platforms

<3 months

Groq Docker Gemini Render SQLite Vercel ChromaDB Firebase Supabase PostgreSQL GitHub Actions Google Cloud Platform Amazon Web Services AWS Hugging Face Transformers

Projects

EcoRAG Agent – Environmental RAG System

LangGraph, ChromaDB, Groq Llama 3.1, Hugging Face

Engineered a fault-tolerant Retrieval-Augmented Generation system with automatic multi-key failover. Built a 4-stage LangGraph agentic pipeline using MiniLM embeddings and ChromaDB, and optimized CPU-only deployment for low-latency inference and fast cold starts.
Green Minds – AI Wellness Journal

Streamlit, Firebase, Hugging Face, Groq, Gemini

Fine-tuned BERTweet with GoEmotions for multi-label emotion recognition, achieving 92% F1-score. Built AI workflows using Groq Llama 3 and Gemini APIs with prompt engineering, and architected a full-stack AI platform with journaling, mood analysis, persistent user data, and real-time AI feedback.

Explore the Broxer community

💬 Q&A Ask about roles, salaries and interviews — and help others. 👥 Circles Join discussions on skills, companies and industries. 📰 Feed Follow updates and insights from the community.