Lead R&D projects to explore, prototype, and implement cutting-edge techniques in LLMs and

conversational AI.

● Focus areas include: advanced RAG strategies, model fine-tuning, agentic workflows, multi-agent

orchestration, and prompt optimization.

● Stay constantly apprised of the latest academic research (e.g., from ACL, NeurIPS, EMNLP) and

apply relevant findings to our platform.

Design, build, and maintain a robust evaluation framework to rigorously benchmark our AI

models and agent performance against internal and external standards.

● Define key metrics for quality, accuracy, latency, cost, and safety.

● Analyze model failure modes in-depth and develop strategies for continuous improvement.

● Own the strategy for our model stack, including when to use proprietary models (e.g., GPT-4o,

Claude 3), open-source models, or fine-tuned specialist models.

● Conduct experiments and lead initiatives for fine-tuning or (P)EFT-tuning models for specific

customer domains and tasks.

● Act as the primary technical consultant to the Product team on AI/LLM capabilities, helping to

define a feasible and ambitious product roadmap.

● Partner with Engineering to develop MLOps pipelines for deploying, monitoring, and updating

models and AI systems in production.

● Provide deep technical guidance and mentorship to our Forward Deployed Engineering team on

complex AI challenges.

● Mentor junior data scientists and engineers on the team, fostering a culture of scientific rigor

and innovation.

● Act as ConvoZen AI's internal and external subject matter expert on data science for generative

AI.

Preferred Qualifications:

Bachelors or MS or PhD in Computer Science, AI, NLP, Machine Learning, or a related

quantitative field.

● 7+ years of hands-on industry experience in data science, with at least 4+ years focused

specifically on NLP and Machine Learning.

● Demonstrable, deep, hands-on experience with the modern LLM stack. You must have:

● Experience building and optimizing complex RAG (Retrieval-Augmented Generation) pipelines.

● Practical experience with model fine-tuning (e.g., using Hugging Face, Axolotl) and

parameter-efficient tuning (P)EFT techniques.

● Deep understanding of agentic frameworks (e.g., LangChain, LlamaIndex) and the principles of

tool use and multi-step reasoning.

● A strong grasp of core NLP concepts, including transformers, embeddings, tokenization, and

dialogue systems.

● Expert-level proficiency in Python.

● Deep experience with modern ML/NLP libraries (e.g., PyTorch, Hugging Face transformers,

datasets, evaluate, LangChain).

● Experience with vector databases (e.g., Pinecone, Weaviate, Milvus).

● A proven ability to run structured experiments, work with ambiguity, and translate complex

research into practical, high-impact solutions.