Responsibilities :
- End-to-end development of the AI stack : from model selection / prompting to formula generation.
- Design and fine-tune LLM-based pipelines (OpenAI, Claude, Code LLaMA, etc.).
- Semantic parsing and logic modelling of financial queries.
- Build internal tools for testing, prompt management, and accuracy tracking.
- Collaborate with product + design to integrate AI into a spreadsheet UI.
The core requirements for the job include the following :
LLMs and Prompt Engineering :
Experience with GPT-4 / Claude / open-source LLMs.Hands-on with LangChain / LlamaIndex.Few-shot prompting, function calling, or fine-tuning.Formula and Code Generation :
Built or worked with text-to-SQL, codegen.Understands formula parsers / DSLs / semantic tree generation.Financial Logic (nice to have) :
Familiar with financial statements (P&L, cash flow).Can reason about business metrics like margin, growth, and cost drivers.ML Engineering and Infra :
Python, FastAPI, embeddings, vector DBs (e.g., FAISS, PGVector).MLOps basics for evaluation, monitoring, and versioning.Bonus Points :
Experience working with Excel APIs or spreadsheet-like UIs (e.g., Handsontable).Exposure to structured finance data (ERP exports, MIS reports, etc.).Built internal tools or prototypes at an early-stage startup.(ref : hirist.tech)