Manish B.

Hi, I'm Manish I'm an AI/ML engineer with over 5 years of experience in building AI/ML products. For the past several years, I’ve worked as both an ML Engineer and Researcher. As an ML Engineer, I design and implement AI agents and RAG systems using modern, production-ready techniques such as: 📌 Embedding models + vector databases for context retrieval, text-to-speech models like Whisper 🔧 Function calling for reliable tool usage 🤖 Agent architectures like ReAct and CodeAct ⚙️ Backend APIs using FastAPI or Node.js What I can help you build 🚀 Production-ready MVPs using LLMs, RAG pipelines, and AI agents from idea to deployment 🧠 Research-driven AI solutions for new products, including model selection, evaluation, and experimentation 🤖 Custom AI agents (coding, research, content, automation) using modern agent frameworks 🔧 Fine-tuned open-source LLMs (SFT / RLHF / RLVR) hosted privately or on-prem ⚙️ Scalable backends & APIs (FastAPI / Node.js) to integrate AI into real systems Libraries/Platforms I have experience with. - OpenAI's GPT models, Claude's models, text-embedding models, text-to-speech models like whisper, sentence-transformers - Langchain, Langraph, Llamaindex - Pytorch, vLLM, SGLang, Huggingface transformers, trl, veRL - Some other relevant libraries such as Scikit-learn, NLTK, NumPy, Pandas, Matplotlib Now here are some of the apps I've developed. 1 PrimeClip.pro A chat-based video editing agent that can edit your videos autonomously, you give the prompt and the agent edits your videos by completing tasks like, adding animations, removing silences, removing filler words/bad takes. The agent used in this project was completely developed from scratch, and uses claude model under the hood to carry out tasks. 2 AI Writing Assistant It is an AI assistant that assists you in your writing, for instance, it can automatically curate detailed, deep researched report by surfing the internet and reading the documents provided (PDF, CSV, DOCX). 3 AI Shorts Generator It creates shorts/reels from a long-form video such as podcast. It analyzes the best hooks from the whole video, cuts them into different parts, rates the best hooks in increasing order and selects the best one for the shorts creation. It also adds automatic transition, subtitles, background music, AI images and videos in between. Now As a Researcher, I have deep understanding of 🧪 LLMs / Transformer models 🎯 Model fine-tuning via SFT, RLHF, and RLVR 📊 Evaluation, experimentation, and optimization My contributions to the field of Natural Language Processing (NLP) have been recognized through a research paper titled "Nepali Encoder Transformers: An Analysis of Auto Encoding Transformer Language Models for Nepali Text Classification” published at SIGUL2022 @LREC2022, held in Marseille, France. Keywords: AI agent, AI developer, AI Automation, Langchain, Langraph, Python Developer, Hermes, OpenClaw