Qureos Mentor

APNA

<div><div>Job Title 
Senior Data Scientist / Research Scientist &mdash; LLM Training &amp; Fine-tuning (Indian Languages, Tool Calling, Speed) 
Location: Bangalore 
 
About the Role 
 
We&rsquo;re looking for a hands-on Data Scientist / Research Scientist who can fine-tune and train open-source LLMs end-to-end&mdash;not just run LoRA scripts. You&rsquo;ll own model improvement for Indian languages + code-switching (Hinglish, etc.), instruction following, and reliable tool/function calling, with a strong focus on latency, throughput, and production deployability. 
This is a builder role: you&rsquo;ll take models from research experiments evals production. 
 
What You&rsquo;ll Do (Responsibilities) 
 
<ul><li>Train and fine-tune open LLMs (continued pretraining, SFT, preference optimization like DPO/IPO/ORPO, reward modeling if needed) for:</li></ul>
Indian languages + multilingual / code-switching 
Strong instruction following 
Reliable tool/function calling (structured JSON, function schemas, deterministic outputs) 
<ul><li>Build data pipelines for high-quality training corpora:</li></ul>
Instruction datasets, tool-call traces, multilingual data, synthetic data generation 
De-duplication, contamination control, quality filtering, safety filtering 
<ul><li>Develop evaluation frameworks and dashboards:</li></ul>
Offline + online evals, regression testing 
Tool-calling accuracy, format validity, multilingual benchmarks, latency/cost metrics 
<ul><li>Optimize models for speed and serving:</li></ul>
Quantization (AWQ/GPTQ/bnb), distillation, speculative decoding, KV-cache optimizations 
Serve via vLLM/TGI/TensorRT-LLM/ONNX where appropriate 
<ul><li>Improve alignment and reliability:</li></ul>
Reduce hallucinations, improve refusal behavior, enforce structured outputs 
Prompting + training strategies for robust compliance and guardrails 
<ul><li>Collaborate with engineering to ship:</li></ul>
Model packaging, CI for evals, A/B testing, monitoring drift and quality 
<ul><li>Contribute research:</li></ul>
Read papers, propose experiments, publish internal notes, and turn ideas into measurable gains 
 
What We&rsquo;re Looking For (Qualifications) 
 
Must-Have 
 
<ul><li>4 - 6 years in ML/DS, with direct LLM training/fine-tuning experience</li><li>Demonstrated ability to run end-to-end model improvement:</li></ul>
data training eval deployment constraints iteration 
<ul><li>Strong practical knowledge of:</li></ul>
Transformers, tokenization, multilingual modeling 
Fine-tuning methods: LoRA/QLoRA, full fine-tune, continued pretraining 
Alignment: SFT, DPO/IPO/ORPO (and when to use what) 
<ul><li>Experience building or improving tool/function calling and structured output reliability</li><li>Strong coding skills in Python, deep familiarity with PyTorch</li><li>Comfortable with distributed training and GPU stacks:</li></ul>
DeepSpeed / FSDP, Accelerate, multi-GPU/multi-node workflows 
<ul><li>Solid ML fundamentals: optimization, regularization, scaling laws intuition, error analysis</li></ul> 

Nice-to-Have 
 
<ul><li>Research background: MS/PhD or publications / strong applied research track record</li><li>Experience with Indian language NLP:</li></ul>
Indic scripts, transliteration, normalization, code-mixing, ASR/TTS text quirks 
<ul><li>Experience with pretraining from scratch or large-scale continued pretraining</li><li>Practical knowledge of serving:</li></ul>
vLLM / TGI / TensorRT-LLM, quantization + calibration, profiling 
<ul><li>Experience with data governance: privacy, PII redaction, dataset documentation</li></ul>
Tech Stack (Typical)</div> 
<ul>
<li>
PyTorch, Hugging Face Transformers/Datasets, Accelerate</li>
<li>DeepSpeed / FSDP, PEFT (LoRA/QLoRA)</li>
<li>Weights &amp; Biases / MLflow</li>
<li>vLLM / TGI / TensorRT-LLM</li>
<li>Ray / Airflow / Spark (optional), Docker/Kubernetes</li>
<li>Vector DB / RAG stack familiarity is a plus</li></ul><div> 

What Success Looks Like (90&ndash;180 Days) 
 
<ul><li>Ship a fine-tuned open model that measurably improves:</li></ul>
Instruction following and tool calling correctness 
Indic language performance + code-switching robustness 
Lower latency / higher throughput at equal quality 
<ul><li>Stand up a repeatable pipeline:</li></ul>
dataset versioning, training recipes, eval harness, regression gates 
<ul><li>Build a roadmap for next upgrades (distillation, preference tuning, multilingual expansion)</li></ul> 

Interview Process</div> 
<ul>
<li>
30-min intro + role fit</li>
<li>Technical deep dive: prior LLM work (training/evals/production constraints)</li>
<li>Take-home or live exercise: design an LLM fine-tuning + eval plan for tool calling + Indic language</li>
<li>Systems round: training/serving tradeoffs, cost/latency, failure modes</li>
<li>Culture + collaboration round</li></ul></div>

Senior Data Scientist / Research Scientist - LLM Training & Fine-tuning

Join APNA and explore 1   exciting career opportunities. They're hiring for various roles with options for remote or on-site work. Apply today and become part of their innovative team.

APNA

ABOUT_COMPANY

Senior Data Scientist / Research Scientist - LLM Training & Fine-tuning