Omkar Thawal

Built an AI pipeline to extract menu items from hotel menu-card photos

Freelance AI Project at Restronear.me (2025-01 – Present)

Built an AI pipeline to extract menu items from hotel menu-card photos

• Built an AI pipeline: image upload → OCR + image processing → LLM-based dish structuring → digital menu generation.
• Extracted dish names, descriptions, prices, and categories from noisy menu photos using open-source OCR + LLM prompt engineering.
• Stack: Python, OpenCV, Tesseract OCR, LLMs (open-source), REST APIs.

Founder & Architect at Tesserant AI (2025-01 – Present)

AI Software Startup (business automation, AI agents, CRM, healthcare AI)

• Founded an AI startup offering business automation, AI agents, CRM automation, healthcare AI, business dashboards, and custom AI applications.
• Designed business model, service offerings, software architecture, and AI workflows across 8+ service categories.
• Built MVP using AI-assisted development (Claude Code, opencode, Codex): Next.js, React, Tailwind CSS, TypeScript, Node.js, Fastify, CockroachDB, Supabase, Cloudflare, Vercel.
• Integrated AI website assistant (chatbot + knowledge base) for prospect-facing service discovery.

AI Clinical Data Extraction Assistant at Client Healthcare Project (2025-01 – Present)

Speech-to-text + LLM structuring (Referral)

• Built pipeline: doctor–patient conversation → Speech-to-Text → LLM summarisation → structured field extraction → multi-column Excel rows.
• Integrated open-source speech-to-text with LLM APIs to convert spoken conversations into semi-structured data.
• Authored and iterated prompts for consistent structured patient notes (chief complaint, history, diagnosis, advice, etc.).
• Stack: Python, open-source STT, LLMs, REST APIs, Excel automation.

Hospital Management System (HIMS) at Independent Project (2024-01 – Present)

In Progress - Architected a scalable Hospital Information Management System

• Architected a scalable Hospital Information Management System (HIMS) with modules for patient registration, appointments, EMR, billing, pharmacy, laboratory, CRM, and analytics.
• Designed a modular, cloud-native backend using Node.js, Fastify, REST APIs, and CockroachDB to support scalable healthcare workflows and multi-hospital deployment.
• Developed the application using Next.js, React, CockroachDB, Railway, Vercel, and Cloudflare, emphasizing modular architecture, role-based access control, and maintainable system design.

Algorithmic Trading Strategy Collaboration at Remote Collaborative Project (2024-01 – 2025-06)

Approximately 6 months - with Canadian derivatives strategist & Chinese app developer

• Collaborated with a Canadian derivatives strategist (15+ yrs experience) and a Chinese app developer to implement two algorithmic trading strategies from 15–30-page strategy documents.
• Translated strategy documents into working Python/MQL5 code: Smart Money Concepts, grid trading, zone recovery, risk management.
• Contributed AI/ML expertise for backtesting, parameter tuning, and trade visualisation.
• Outcome: one strategy underperformed in live market conditions; the second delivered modest returns (truthfully reported).
• Stack: Python, Pine Script, REST APIs, MQL5, MetaTrader 5.

Artificial Intelligence & Machine Learning Intern at AB Infotech Solution (2024-01 – 2024-05)

• Contributed to 5+ real-world data science projects focusing on data cleaning, feature engineering, and model deployment using Pandas, NumPy, scikit-learn, Matplotlib, Keras, PyTorch.
• Designed and optimised deep learning models (CNN, RNN, Transformer, Reinforcement Learning), achieving up to 92% accuracy and reducing error rates by 30%.

BE in Artificial Intelligence & Data Science – Ajeenkya DY Patil School of Engineering (2021-07 – 2025-06)

– New English School And Jr College Of Science (2019-05 – 2021-03)

– Shri MulikaDevi Vidyalaya (2018-06 – 2019-03)