Lead / Staff Data Scientist – Core LLM & Model Fine-Tuning

Location

Bangalore, Karnataka, India

Salary

Not specified

Company

Akash pawar

Job description

• Role & Company / Project Overview: Oneshot Consulting Pvt. Ltd. is hiring exceptional Lead / Staff Data Scientists with hands-on expertise in LLM training, fine-tuning, and model optimization. This opportunity is specifically for professionals who have worked directly on improving the models themselves, rather than just building applications around existing LLMs. • Key Responsibilities & Daily Duties: - Train, fine-tune, and optimize Large Language Models (LLMs) from the ground up and at the model architecture/weight level. - Fine-tune specific foundation and open-source models such as Llama, Mistral, Qwen, BERT, T5, Phi, or Mixtral. - Implement modern tuning methodologies including LoRA, QLoRA, PEFT, Supervised Fine-Tuning (SFT), DPO, or RLHF. - Execute fine-tuning projects with specific datasets, rigorous implementation details, and measurable model-level evaluation metrics (accuracy, F1-score, precision/recall, perplexity). - Drive core LLM improvements, pretraining, training from scratch, model distillation, or quantization-aware training. • Required Qualifications, Skills & Experience: - Mandatory Core LLM Expertise: Proven, hands-on experience in training and fine-tuning Large Language Models (LLMs) within the last 1–2 years strongly preferred. - Professional Experience: Minimum 6+ to 7+ years of relevant professional experience. - Educational Background: Preference for candidates from IIT, IISc, BITS, or equivalent top international institutions (including leading US/EU universities). Educational flexibility is possible for exceptionally strong candidates. Research experience or a PhD is a strong advantage. - Team Background: Experience in AI-native startups, research labs, or ML platform teams where model training is a core responsibility is highly desirable. - Additional Technical Differentiators: Experience with model pretraining, training models from scratch, model distillation, quantization-aware training, or research publications at top NLP/ML venues (ACL, EMNLP, NAACL, NeurIPS, ICML, ICLR). - Critical Screening Criteria: Candidates focused primarily on RAG pipelines (LangChain, LangGraph), prompt engineering, agentic AI (CrewAI, AutoGen, MCP), integrating/consuming third-party APIs (GPT-4, Claude), or classical ML/CV backgrounds without demonstrable model fine-tuning will not be considered. • Working Hours, Shifts & Location Details: - Location: Bangalore, Karnataka, India (Preferred). - Work Mode: Hybrid. • Compensation & Benefits: - Competitive industry package commensurate with experience. • Application Instructions: - Interested candidates and relevant referrals can share their updated profiles with the recruitment team.