Open to opportunities

Chappidi Preethi

@chappidipreethi

Message

AI engineer and ML/vision builder, productionizing RAG and fraud-detection pipelines with measurable accuracy gains.

India

Message

What I'm looking for

I’m looking for a role where I can productionize ML: build RAG/NLP and computer-vision systems, improve real-time accuracy, and ship end-to-end pipelines. I want ownership, fast iteration, and measurable impact.

I’m an “AI Engineer” focused on turning applied ML into reliable, production-ready systems. In my recent roles, I’ve improved throughput, reduced repeated outputs, and built end-to-end pipelines that personalize results using real user context.

At SproutsAI, I built and productionized a job-matching pipeline using JD–resume embeddings (Qdrant cosine-similarity ranking plus geo-filtering) and increased interview fraud-detection accuracy by integrating video-derived eye-pupil movements. I also worked on async prompt optimization to boost question-generation throughput by 25× while eliminating ~50% repetition.

I complement ML with computer vision and real-time intelligence—enhancing multi-object tracking by integrating BoT-SORT/Re-ID with YOLOv8 and strengthening the quality of tracking and event detection using InpaintNet and background subtraction. I’m excited to keep building systems that combine RAG, deep learning, and vision for practical, user-facing impact.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning Intern

Current

SproutsAI

Jul 2025 - Present (11 months)

Boosted question-generation throughput by 25× and reduced ~50% question repetition via async processing and prompt optimization. Built and productionized a job-matching pipeline using JD–resume embeddings (Qdrant cosine-similarity ranking + geo filtering) and an interview fraud-detection pipeline; improved audio reading/natural detection accuracy from 29% to 58.6%.

Sentence JD Resume Embeddings Qdrant Cosine Similarity Ranking Geo Coordinate Filtering MediaPipe WavLM Embeddings Fraud Detection Mediapipe

Current

AI Engineer

Current

THP IIT Bombay

Jan 2025 - Present (1 year 5 months)

Enhanced real-time multi-object tracking by combining BoT-SORT, Re-ID, and a YOLOv8 detector for consistent ID tracking. Built a RAG chatbot with LangChain and LLaMA 2 (7B) featuring memory, multi-hop retrieval, reranking, web search (DuckDuckGo API), and TTS responses; also developed a stereo disparity/block-matching depth estimation model for distance estimation and safer robot navigation improv

YOLOv8 Retrieval Augmented Generation (RAG)Lang Chain LLaMA 2 Reranking TTS

Computer Vision Intern

Visist.AI

Jan 2025 - Mar 2025 (2 months)

Improved shuttle tracking reliability by analyzing detection gaps and addressing motion blur using InpaintNet with temporal context. Enhanced shuttle fall detection near white courtlines by isolating the shuttle via background subtraction.

Root Cause Analysis Shuttle Tracking Motion Blur