ADITYA KANCHI
Building scalable production architectures and exploring the boundaries of AI safety & research.
Featured Projects & Research
Production-grade AI architectures, hackathon winners, and open-source foundation models.
FinArtha
Advanced financial tracking assistant utilizing Streamlit, Groq Llama 3.3 70B, and IBM Granite Open Source models running on Hugging Face Spaces for real-time portfolio analysis and risk evaluation.
Snowy (v1 & v2)
Intelligent project tutor and repository evaluation agent. v1 utilizes Groq Llama 3.3 70B, Streamlit, Supabase, and GitHub Auth. v2 scales the setup entirely with a high-performance frontend framework, robust backend infrastructure, and a custom RAG pipeline.
Bhasha Mitra
Open-source regional language assistant built leveraging Streamlit alongside open-source models hosted directly via Hugging Face. Facilitates multi-lingual accessibility, local dialect processing, and translation.
DeltaTwo
A production-grade startup ecosystem focused strictly on AI research alignment, safety protocols, and community integration. Currently advancing through its v0 launch phase to empower builders.
Engineering Against the Grain
Blending AI/ML technical rigor with philosophy, cognitive psychology, and production-grade startup building.
AI/ML Engineering & Philosophy
Holding a formal specialization in Artificial Intelligence and Machine Learning, my work is rooted in complex problem-solving, AI safety research, and the intersection of cognitive psychology and philosophy.
Emerging from an underdog, less-reputable institution, I turned structural constraints into an unfair advantage—building relentless self-reliance and raw engineering velocity. Rather than waiting for traditional validation, I focused on winning major national hackathons and turning deep technical research into production-scale startup architectures.
My philosophy centers on building safe, resilient systems: engineering artificial intelligence that expands human capability while adhering strictly to model alignment, safety protocols, and transparent architecture.
Skill Capsule Matrix
Scaling Snowy & DeltaTwo
FinArtha Victory
IBM Hackathon 1st PlaceEngineered FinArtha under high-pressure competitive constraints, capturing 1st place out of 1,400+ participants with Groq Llama 3.3 70B and IBM Granite OS.
Snowy Execution
Production Tutor & Git EvalTransformed hackathon concepts into a full-scale automated tutor & Git evaluation engine (v1). Scaled v2 with a high-performance frontend framework, robust backend infrastructure, and custom RAG pipeline.
DeltaTwo Ecosystem
AI Safety & AlignmentFounded DeltaTwo as a production-grade startup ecosystem focused strictly on AI safety protocols, model alignment research, and connecting researchers to build human-aligned AI architectures.