Real systems, documented design decisions, and the evaluations behind them.
Local-first autonomous desktop agent built around persistent memory, document RAG, tool use, sandboxed execution, human-in-the-loop approvals, and configurable multi-provider LLM routing.
Read case studyThree connected modes: cited course-material chat, an academic consultation/research agent with plan approval and editable questions, and professor-controlled exam-paper creation.
Read case studyOffline RAG over a secondary-level physics textbook: two dense retrieval paths, FAISS indexing, candidate merging, cross-encoder re-ranking, relevance filtering, and follow-up question reformulation with local Gemma 2 via Ollama.
Read case studyHierarchy-aware embedding model that represents inputs from coarse intent to fine-grained meaning via a frozen Transformer backbone, learned token-attention pooling, and iterative hierarchical refinement.
Read case studyText-and-voice conversational screening support that tracks which ADHD symptom indicators are covered — no fixed questionnaire flow.
Read case studyPost-training pruning toolkit supporting magnitude, Wanda, SparseGPT, RIA, and an activation-aware ClearCut criterion, with unstructured and hardware-oriented N:M sparsity.
Read case studyFull ML pipeline over Rawalpindi road traffic accidents targeting injury severity and post-accident patient status.
Read case study