Event-driven clinical-trial patient safety platform

An AI-powered clinical-trial safety platform that bridges the gap between in-person visits. Combines real-time biometric anomaly detection, multi-agent voice/text check-ins, and researcher triage — normalizing symptoms into MedDRA terms and CTCAE severity grades with a human in the loop.

AI financial twin & real-time shopping companion

A personal-finance platform with a multimodal receipt-OCR pipeline (GPT-4o Vision), a pgvector-backed spending memory, and a Chrome extension that computes the 'true cost' of a purchase in hours-of-work while you shop.

Intelligent LLM context-compression engine

A high-performance LLM input-compression framework. A fine-tuned BERT token classifier scores every token by semantic necessity, then a two-tier pruning engine (context + token level) cuts prompt size while a zero-hallucination constraint engine preserves code and structure.

Distributed LLM inference engine in C++20

A distributed LLM inference engine that runs models like Llama 3 70B across a ring of heterogeneous consumer devices using pipeline parallelism — bypassing the VRAM wall. Hand-written AVX2/NEON SIMD kernels, a zero-copy Linux kernel-module transport, and a custom INT8/FP32 weight serializer.

Encrypted ML inference engine (Rust + Python)

A privacy-preserving ML inference library that runs PyTorch / sklearn / XGBoost models under Fully Homomorphic Encryption — inputs and outputs never leave ciphertext. A 'narrow-waist' 3-layer architecture lowers ONNX graphs to a versioned IR executed on tfhe-rs primitives, with a PyO3 bridge and bit-for-bit exactness guarantees.

Autonomous agentic AI SDR platform

An end-to-end agentic sales-development platform. Multi-agent workflows discover leads, run multimodal visual audits of their sites (Gemini Vision + Playwright), auto-generate improved website prototypes, and reach out over cold email and real-time bi-directional AI voice calls.