ML Engineer

Summary

I'm researching a self learning conversational agent that uses a custom knowledge graph to store knowledge triplets extracted from conversations to reduce hallucinations and ground the LLM. The knowledge graph uses a word embedding model to do semantic deduplication over new triplets. I've implemented multi-hop function calling using structured output parsing from scratch and is capable of solving GSM8K problems and HotPotQA.

Enhanced multi-step reasoning capabilities, achieving a 35% increase in correct answers over GSM8K and
HotPotQA datasets, by integrating ReAct and Graph-of-thought for advanced multi-hop functionality.
Developed a custom chatbot backend, serving as the core platform functionality, which streamlined prompt management and conversation storage, utilizing the OpenAI GPT API for dynamic response generation.
Designed and implemented a knowledge graph schema to optimize semantic match accuracy, employing vector embeddings of knowledge triplets with notably improved semantic alignment.

Expectations

Working with AI or large language models. Can do backend engineering with databases and APIs. Also know React.js for frontend development.

Employment Preferences

Expected Base Salary

**,000 USD

Academic Degree

Experience

Skills

• Python
• C++
• MIPS Assembly
• Solidity
• Tensorflow
• Flask
• React.js
• LangChain
• OpenAI