Projects

I'm working on hosting these projects here. Full write-ups and demos are currently under construction.

Tesera — Open-Source Media Verification Under construction

An open-source SDK that combats AI-generated misinformation by creating tamper-proof cryptographic records for images and videos — tracking who created a file, how it was modified, and verifying authenticity. Edit history is modelled as a directed acyclic graph with BFS for chain traversal and ancestor detection.

PythonSQLiteGitCryptographic signing
Modular Synthetic Data & Finetuning Under construction

A modular synthetic data generator for fine-tuning open-weight LLMs on domain-specific business workflows. Includes an LLM-as-judge evaluation loop (via LangChain) to score, filter, and improve synthetic data before tuning. Fine-tuned Llama on payment processor ToS to produce a compliance assistant.

PythonClaude APILangChainLoRALlama
AI Asset Manager & Tax Loss Harvesting Under construction

An asset manager that lets users describe portfolio weights and risk parameters in natural language. Handles daily portfolio reconstruction and tax-aware rebalancing, with a LangChain/FAISS RAG layer that explains notional and lot-level discrepancies in plain English.

PythonOpenAI APISQLPandasFAISS