Prashanth
StarOne01
Curiosity-driven engineer exploring systems from first principles. Building high-performance distributed systems, AI voice agents, and production ML pipelines.
About
I am a Systems and AI Engineer based in Coimbatore, Tamil Nadu. Currently a third-year Electrical & Electronics Engineering student, I am a self-taught programmer driven by an insatiable curiosity about how complex systems work. I specialize in building software from zero-to-one through deep exploration and investigative problem-solving.
As the Founding Engineer at Medclara, I architect voice-first, multilingual clinical documentation platforms that convert doctor-patient conversations into structured SOAP notes using custom ASR and LLMs. I also run a Software Development Agency delivering high-end enterprise software and AI agent systems for diverse businesses, while building an upcoming startup called primesoma.
My core expertise spans production machine learning pipelines running local inference with fine-tuned Whisper and Indic models, down to core compiler infrastructure with over 15 merged PRs to LLVM. I bridge the complex constraints of local bare-metal inference and responsive UI to deliver optimal, robust solutions.
Expertise
Production Machine Learning
Developing state-of-the-art ASR architectures (Conformer, RNN-T, CTC). Fine-tuning Whisper models with QLoRA, handling speaker diarization, and deploying local inference pipelines (Ollama) with structured clinical outputs.
Systems & Infrastructure
Architecting ABDM/ABHA, FHIR R4 integrations and targeting rigorous Indian compliance standards (DPDP Act). Building on-premise PostgreSQL architectures and deterministic concurrent Go/C++ backends.
Open Source & Compilers
Active LLVM contributor with 15+ merged PRs including float-related changes. Creator of PhraseNuX, a C++ CLI password manager with AES encryption, and researcher on cross-lingual reasoning compression.
Creative Technologies
Fusing technical rigor with high-end creative tools, working with Unreal Engine 5 cinematics and DaVinci Resolve for advanced color grading and export pipelines.
Experience
Founding Engineer
PresentMedclara
Architecting a voice-first, multilingual clinical documentation platform. Built a pipeline achieving ~88/100 clinical accuracy from raw Tamil inputs in a single call. Integrated ABDM/FHIR R4 targeting Tier-2/3 cities with on-premise deployments.
Founder
PresentSoftware Development Agency
Running a Coimbatore-based agency funding other ventures. Delivering high-end web projects, custom AI agent systems (like WhatsApp bots), and enterprise software for real estate firms, studios, and manufacturing businesses.
Founder
Upcoming Startupprimesoma
Building a vertically integrated wellness ecosystem — everything a prime body needs from food and groceries to farms, fitness, and health. Owning the entire stack to deliver seamless, end-to-end wellness solutions.
Open Source Contributor & Researcher
OngoingLLVM, EfficientXLang
Authored 15+ merged LLVM PRs. Conducted original research assessing cross-lingual reasoning in multilingual LLMs. Created PhraseNuX (C++ CLI password manager) and authored technical series on production AI realities.
Technical Stack
Languages & Frameworks
AI & ML Systems
Production & Infrastructure
Low-level & Compilers
Creative & Motion
Contact
I welcome discussions on AI technology, systems architecture, engineering leadership, and opportunities to collaborate on meaningful projects pushing the boundaries of what's possible.
Recent Research & Work
- →15+ Merged LLVM PRs (float-related changes)
- →EfficientXLang: Cross-lingual reasoning gap in LLMs
- →PhraseNuX: C++ CLI AES password manager
- →Active content series: Multilingual LLM evaluation & ASR architecture