Job Description

Model Trainer (Post-Training and Evals)

n

The job

n

Take an agentic task that currently runs on a frontier model, fine-tune an open-weight model on it, and prove with evals that the small model does the task better and cheaper. Then do it again on the next task, and the next.

n

You will work on real enterprise workloads. Some of it runs on customer GPUs, some on ours. You own the model from data through to the eval report that convinces a customer to switch traffic over.

n

What you will do

n

n
Build training sets from real agent traffic and traces, not scraped benchmarks
n
Run LoRA, QLoRA, DoRA, full fine-tunes, CPT, DPO and GRPO depending on what the task needs
n
Design task-specific evals for agentic behavior: tool calling accuracy, multi-step reliability, output format adherence, groundedness. Benchmarks are not the bar, the customer's task is
n




Get the model serving on vLLM and measure real latency and token cost against the frontier baseline
n
Work inside ShadowLM, our open-source fine-tuning SDK, and improve it as you go
n

n

What we are looking for

n

n
You have post-trained open-weight models yourself and can talk about what failed, not just what worked
n
Comfortable across llama, qwen, mistral, gemma, phi, deepseek and whatever ships next month
n
Robust on evals. You know that most fine-tuning projects die because nobody built an honest measurement first
n
You can read a paper on Monday and have it running on Wednesday
n
Practical about hardware. You know what fits on one GPU and what does not
n

n

Nice to have

n

n
Reward modeling or RL for agents
n
Quantization and distillation
n
Experience getting a model past an enterprise security review
n

📌 Model Trainer Post Training And Evals Bengaluru
🏢 Lyzr AI
📍 Bengaluru