Twenty builds across LLM systems, RAG, computer vision, automation, voice AI, quantitative trading and governance tooling.

LLM observability & safety platform: model registry with risk classification, audit logging, PII & toxicity detection, cost tracking across 250+ models, NIST AI RMF compliance reports, published Python SDK.

RAG system letting legal teams query large PDF documents in natural language with source-cited answers. Supports Gemini, OpenAI and local Ollama backends.

Production outbound and inbound voice agent for automated phone qualification and warm transfer. Multi-tenant, TCPA-compliant and backed by 586 hermetic tests, with a concurrency-safe dispatcher and warm-transfer routing. Built for high call volume at ~$0.10-0.15/min.

AI knowledge system that ingests web and video links, scrapes and stores the content, and answers user questions over it with RAG. Built at Viral Tech.

Real-time recognition via webcam: 1,662-dimensional MediaPipe Holistic keypoints classified by a CNN-LSTM, rendered as Urdu text and speech. Final year project.

Visibility enhancement in foggy driving conditions: dark channel prior dehazing + CLAHE contrast, feeding YOLOv8 detection of pedestrians, vehicles and traffic signals.

Real-time vehicle license plate detection and number reading using YOLOv8, designed for traffic-signal monitoring applications.

Automated attendance with real-time facial recognition, Firebase integration, live video feedback and downloadable reports.

Deep learning system that interprets images and answers natural-language questions, combining computer vision and NLP techniques.

Offline backtesting harness validating algorithmic trading strategies over ~9 years of BTC, ETH and SOL data, net of realistic fees and slippage. Enforced strict out-of-sample discipline (no-look-ahead, verified by tests), proved the strategies underperformed buy-and-hold, and found a risk overlay that cut max drawdown ~27%. Shipped 43 hermetic tests and a go/no-go report.

AI resume builder that tailors a resume to a job description, scores ATS compatibility, and formats it to pass applicant tracking systems.

Recruitment automation that generates screening criteria from a job spec, pulls resumes via email, ranks candidates with AI matching, and sends bulk templated outreach.

Unified interface over GPT-4, Claude, Gemini and LLaMA with model comparison, switching and performance analytics for text, chat and code tasks.

Multimodal medical assistant that analyzes medical images, answers health questions, and supports voice interaction using LLMs and computer vision.

Touchless calculator: an on-screen keypad operated by finger-pinch detection through a live webcam feed for real-time arithmetic.

Streamlit app that turns natural-language questions into SQL with Groq's LLaMA 3, querying a live SQLite database and returning results in an intuitive UI.

Pipeline that turns YouTube videos into blog posts and podcasts: fetches transcripts, summarizes with AI, generates audio, and ships to Drive and Sheets.

Django system that generates location-specific, Axios-style newsletters with Claude, sourcing fresh articles via Perplexity and structuring them into HTML.

Medical-imaging system that segments arterial plaque in ultrasound images with YOLOv8, annotating regions with confidence scores for cardiovascular diagnostics.

CrewAI multi-agent system that researches the market, plans content, generates visuals and writes captions to automate Instagram content end to end.