Become a sponsor to Jeevesh Singale

Hey there! 👋

I'm Jeevesh Singale (@jeevesh2515), an AI Engineer based in London 🇬🇧.

I build systems where precision, low latency, and zero hallucination are mission-critical. My primary open-source work focuses on evidence-grounded Clinical AI and deterministic developer tooling, alongside building VoxFlow, my commercial conversational voice AI platform.

🏥 Flagship Open-Source Project: Clinical Workflows

🩺 Clinical RAG Agent · Live App Demo

• The Problem: General LLMs routinely hallucinate medical citations, fail at basic clinical calculations (e.g. MAP, BMI, pulse pressure), and lack deterministic guardrails required for patient safety. In healthcare, a single hallucination is catastrophic.
• The Engineering Solution: A production-grade, open-source Clinical Decision Support RAG system designed for zero hallucinations:
  • LangGraph Stateful Safety DAG: Directs queries through strict safety classification, canonical Open Knowledge Format (OKF) facts, and PubMed evidence retrieval before any answer generation.
  • Deterministic Biomarker Engine: Executes clinical formulas via native mathematical validators rather than allowing LLMs to guess calculations.
  • Strict Citation Verification: Every medical claim is verified and cross-referenced against authoritative biomedical literature.
• Validation: Verified with 258 passing tests and evaluated against 55 golden clinical test cases using RAGAS and LangSmith.
• Use Cases: Chronic hypertension management, clinical evidence synthesis, structured triage Q&A, medical research assistance.

🎙️ Commercial SaaS Venture: VoxFlow

⚡ VoxFlow Voice Agent · voxflow.cc

• What it is: My commercial full-duplex conversational voice SaaS product engineered for human-grade real-time voice interactions.
• Core Achievement: Achieved ~200ms glass-to-glass turn latency using Amazon Connect, Groq Whisper STT, Llama-3-70B, and Edge-TTS with speculative interruptible VAD.
• Architecture Rigor: Backed by 952 passing tests, validating real-time packet streaming, echo cancellation, and seamless conversational barge-in under heavy load.

🧠 Knowledge & Developer Tooling

• Cortex: Local-first cognitive memory engine and knowledge graph accelerator for Obsidian vaults and researchers.
• pre-push-verify: Lightweight, zero-overhead git pre-push quality and test harness validator.
• readme-guardian: Automated CI/CD documentation and README consistency validator published on PyPI.

💡 Why Sponsor My Open-Source Work?

Developing zero-hallucination clinical AI systems and maintaining open evaluation benchmarks requires:

• 🔬 Biomedical Benchmark Infrastructure: Continuous evaluation on LangSmith and RAGAS across biomedical datasets and clinical QA corpora.
• ⚡ Open Evaluation Datasets: Publishing reproducible, golden test sets and evaluation benchmarks for the open-source healthcare AI community.
• 🌐 Public Live Demos: Hosting interactive demo applications and cloud sandboxes for developers and clinicians.
• 🛡️ Independent Open Research: Keeping core clinical evaluation frameworks, guardrails, and developer tools 100% open source under the MIT License.

Your sponsorship directly funds cloud compute, API evaluation costs, and dedicated engineering time to advance reliable, zero-hallucination open-source AI.

🏢 Corporate & Healthcare Tech Sponsorship

If your team, startup, or clinic is building clinical AI systems, conversational voice agents, or RAG pipelines:

• Bronze & Silver Sponsors receive your logo and backlink featured on the Clinical RAG Agent documentation and README.
• Gold Sponsors receive direct technical advisory, architecture reviews on RAG hallucination elimination, and guidance on low-latency voice pipelines.

(Reach out via email or GitHub for custom sponsorships and clinical collaborations!)

Thank you for supporting transparent, safe, and evidence-grounded AI! ❤️