Machine Learning Engineer

ConveGenius; Dadri; On-site; INR 1,800,000 - 2,400,000; Full time

Job summary

ConveGenius seeks a specialist to fine-tune education-focused LLMs, owning end-to-end workflows from data curation to evaluation. You will implement LoRA/QLoRA techniques, build RLHF pipelines, and optimize GPU training efficiency across multi-GPU setups.

The role requires deep expertise in PyTorch and Hugging Face, distributed training, and evaluating model performance with robust benchmarks. Experience with open-source and sovereign LLMs is a plus.

Qualifications

• Strong experience in fine-tuning and optimizing large language models (LLMs).
• Hands-on experience with LoRA, QLoRA, SFT, DPO, RLHF, or similar fine-tuning techniques.
• Proficiency in PyTorch and the Hugging Face ecosystem (Transformers, PEFT, TRL).
• Experience with distributed and multi-GPU model training.
• Strong understanding of model performance, evaluation, and optimization.
• Knowledge of DeepSpeed, Megatron-LM, and large-scale training frameworks.
• Understanding of AI/ML pipelines, data preparation, and model deployment.
• Experience working with open-source and sovereign LLMs; focus on standard, production-proven frameworks.

Responsibilities

• Fine-tune foundation models for education-specific tasks: question answering, content generation, adaptive feedback, and curriculum alignment.
• Own end-to-end fine-tuning workflows: dataset curation, training runs, hyperparameter tuning, evaluation, and model versioning.
• Implement efficient fine-tuning methods (LoRA, QLoRA, DoRA, adapters) appropriate to available compute budgets.
• Build RLHF and preference optimisation pipelines: DPO, PPO, and reward modelling for aligning models to learning outcomes.
• Optimise GPU training efficiency: DeepSpeed, FSDP, gradient checkpointing, mixed precision, and multi-GPU setups.
• Evaluate fine-tuned models rigorously: perplexity, task-specific benchmarks, human eval, and regression testing.
• Build data pipelines for instruction-tuning datasets, including multilingual and Indic language data.
• Assess new open-source model releases for domain applicability and adoption readiness.
• Define and track model performance metrics, evaluation benchmarks, and optimisation targets across all fine-tuned model versions.
• Work with open-source and sovereign LLMs, owning the full model adaptation lifecycle using proven, industry-standard frameworks.

Skills

LLM fine-tuning

LoRA/QLoRA/Adapters

PyTorch

Hugging Face

Distributed training

DeepSpeed

Evaluation & benchmarks

Data pipelines

Tools

Transformers

PEFT

TRL

DeepSpeed

Megatron-LM

Job description

Responsibilities

• Fine-tune foundation models for education-specific tasks: question answering, content generation, adaptive feedback, and curriculum alignment.
• Own end-to-end fine-tuning workflows: dataset curation, training runs, hyperparameter tuning, evaluation, and model versioning.
• Implement efficient fine-tuning methods (LoRA, QLoRA, DoRA, adapters) appropriate to available compute budgets.
• Build RLHF and preference optimisation pipelines: DPO, PPO, and reward modelling for aligning models to learning outcomes.
• Optimise GPU training efficiency: DeepSpeed, FSDP, gradient checkpointing, mixed precision, and multi-GPU setups.
• Evaluate fine-tuned models rigorously: perplexity, task-specific benchmarks, human eval, and regression testing.
• Build data pipelines for instruction-tuning datasets, including multilingual and Indic language data.
• Assess new open-source model releases for domain applicability and adoption readiness.
• Define and track model performance metrics, evaluation benchmarks, and optimisation targets across all fine-tuned model versions.
• Work with open-source and sovereign LLMs, owning the full model adaptation lifecycle using proven, industry-standard frameworks.

Requirements

• Strong experience in fine-tuning and optimising large language models (LLMs).
• Hands-on experience with LoRA, QLoRA, SFT, DPO, RLHF, or similar fine-tuning techniques.
• Proficiency in PyTorch and the Hugging Face ecosystem (Transformers, PEFT, TRL).
• Experience with distributed and multi-GPU model training.
• Strong understanding of model performance, evaluation, and optimisation.
• Knowledge of DeepSpeed, Megatron-LM, and large-scale training frameworks.
• Understanding of AI/ML pipelines, data preparation, and model deployment.
• Experience working with open-source and sovereign LLMs; focus on standard, production-proven frameworks.