Prakhar Dwivedi is an AI Engineer at Pear Protocol, where he focuses on building and enhancing production AI systems that facilitate real-world crypto trading workflows. His expertise spans AI agents, LLM integrations, backend services, and distributed systems. He has also interned at Zecruit AI, where he developed a resume parsing pipeline leveraging RAG techniques. Prakhar is currently pursuing a Bachelor of Technology in Electronics and Communication Engineering at IIIT Bhopal, expected to graduate in 2027, and has a strong foundation in software engineering and AI technologies.
- Startups
- Agentic AI
- Web3 / Crypto
- Building with AI
- Engineers
- Investors
- Founders
- Recruiters
- Traning and Placement Point of contact and Coordinator @IIIT Bhopal
- Working on enhancing an AI powered recruitment platform and agentic support.
“PanditAI started as a long-standing idea we finally decided to build. Together with Aditya Sharma and Ishaan Tiwari, we built a full-stack Vedic astrology platform that combines precise astronomical calculations with AI-generated interpretations. I owned the web engineering side, and the hardest part was bridging heavy Python-side computations with a frontend that actually felt usable and engaging. One place I got properly stuck was rendering birth charts generated by backend Python logic. The…”
“got my @EFDevcon ticket! Mumbai it is🇮🇳”
“Started as an idea, found its way into the Best Possible and 10x Impact Labs AI Hackathon, and now has started to take a stronger shape than before. Happy to share that MedBud, a healthcare project I’ve been building on the side, will now be supported by the Ground Truth Fellowship for the next 3 months. With mentorship and support, goal is to take it further, get more users, and shape it into something genuinely useful for the people. I met the amazing team in Dharamshala last month, and it wa…”
“update It now has prompt tuning abilities. 1 - removes filler words and bloat i say. helps with prompting 2 - fills in the gaps and cases i might miss 3 - Directly task based answers 4 - personalised TTS”
“load wise its light. 700mb ram and 1.2gb vram on my 4050, cpu is idle until i press f9 and yea the point is keeping it local. nothing leaves the laptop could go smaller with a distilled model or int8 on cpu if i move it to a separate box, trying that”
“something stuck in my mind last week that in future we wont need to type to interact with machines. wanted to have it setup locally and got pretty good results with a wisper model, tuned to my style. made a script that runs on boot, detects, just pressing f9 and viola.”
FastAPI