Machine Learning Engineer - Model Fine-Tuning & Edge Deployment

Quantiphi

Atlanta (GA)

On-site

USD 140,000 - 190,000

Full time

7 hours ago

Be an early applicant

A complete application in a minute

Tailored resume and cover letter, ready to send.

Get past ATS filters

Job summary

Quantiphi is seeking a senior Machine Learning Engineer to join the AI/ML team in Atlanta. You will own model fine-tuning, evaluation, and production deployment for AI agents and models used in industrial use cases. You will collaborate with AI Architects, Data Scientists, and deployment engineers to deliver end-to-end AI solutions.

You will optimize LLM/SLM agent behavior, build data pipelines, and develop benchmarking frameworks to balance accuracy and latency in real-time applications.

Qualifications

Bachelor's degree in Computer Science, Engineering, Data Science or related.
3–7 years of hands-on ML engineering experience (as per job context).
Experience taking models from experiment to production.

Responsibilities

Fine-tune models (LoRA/QLoRA, RLHF, DPO, SFT) for tool-calling and domain tasks.
Build data pipelines for cleaning, augmentation, and split data for training/validation.
Conduct prompt engineering to optimize agent behavior in production.
Design benchmarks: function-calling accuracy, retrieval precision, latency trade-offs.
Develop end-to-end ML deployment for edge environments (IoT, embedded).
Collaborate with AI Architect, Data Scientists, and Software Engineers.

Skills

LLM fine-tuning

Python

PyTorch

Hugging Face

LLM inference

GPU profiling

Benchmarking

Data pipelines

Deployment & infra

Communication

Education

Bachelor's degree in CS/Engineering/Data Science

Master's or PhD preferred

Tools

Docker

Kubernetes

Linux

NVIDIA Nemotron/NeMo

LangChain / LlamaIndex

MLflow / Weights & Biases

Job description

Quantiphi is an award-winning, AI-First digital engineering and consulting company focused on delivering high-impact Services and Solutions that help organizations solve what truly matters. We partner with enterprises to reimagine their businesses through intelligent, scalable, and transformative AI driving measurable outcomes at the very core of their operations.

Since our founding in 2013, Quantiphi has tackled some of the world’s most complex business challenges by combining deep industry expertise, disciplined cloud and data engineering practices, and cutting‑edge applied AI research. Our work is rooted in delivering accelerated, quantifiable business value, not just technology for technology’s sake.

Headquartered in Boston, Quantiphi is a global organization with 4,000+ professionals serving clients across key industry verticals, including BFSI, Healthcare & Life Sciences, CPG, MFG, TME etc. As an Elite and Premier partner to leading cloud and AI platforms such as NVIDIA, Google Cloud, AWS, and Snowflake, we build and deliver enterprise‑grade AI services and solutions that create real-world impact.

We’ve been recognized with:

21x Google Cloud Partner of the Year awards in the last 8 years.

3x AWS AI/ML award wins.

3x NVIDIA Partner of the Year titles.

2x Snowflake Partner of the Year awards.

We have also garnered top analyst recognitions from Gartner, ISG, and Everest Group.

We offer first‑in‑class industry solutions across Healthcare, Financial Services, Consumer Goods, Manufacturing, and more, powered by cutting‑edge Generative AI and Agentic AI accelerators.

We have been certified as a Great Place to Work for the third year in a row- 2021, 2022, 2023.

Be part of a trailblazing team that’s shaping the future of AI, ML, and cloud innovation.

Your next big opportunity starts here!

Preferred Location: Atlanta, GA

Experience Level: 7-10 Years of Experience

Role Overview -

We are seeking a highly skilled and motivated Machine Learning Engineer to join our AI/ML team. You will join a team that delivers AI agents and models into production for industrial use cases. You will own the path from experiment to production: fine‑tuning, evaluation, inference optimization and benchmarking. You will work with the AI Architect, Data Scientist, Software Deployment and Full Stack engineers, and customer SMEs.

Key Responsibilities

Model fine‑tuning and evaluation -

Fine‑tune Nemotron, Gemma, or any other open‑source models using techniques such as LoRA, QLoRA, RLHF, DPO, and SFT (Supervised Fine‑Tuning), for tasks such as tool calling, RAG and domain‑specific responses.

Build data pipelines for cleaning, curation, synthetic data generation, and train, validation and golden test sets.

Conduct prompt engineering and prompt tuning to optimize LLM/SLM agent behavior for specific use cases

Build evaluation frameworks covering function‑calling accuracy, retrieval precision, task benchmarks and the accuracy‑versus‑latency trade‑off.

Iterate on agent performance through systematic experimentation and hyperparameter optimization

Data Transformation & ML Modeling -

Develop and maintain end‑to‑end data transformation pipelines for preparing training datasets, including data cleaning, augmentation, tokenization, and formatting

Design ML models and training architectures aligned with agent task objectives

Curate and manage high‑quality instruction‑tuning and RLHF datasets for agent training

Apply data‑centric AI methodologies to improve model quality through better data rather than solely model changes

Build and validate benchmarks and evaluation datasets to track model improvements

GPU & Inference optimization, Edge Deployment -

Architect and implement ML model deployment pipelines for edge environments (IoT devices, embedded systems, on‑premise hardware)

Profile and optimize LLM inference for TTFT, end‑to‑end latency, throughput and GPU memory.

Tune vLLM and TensorRT-LLM deployments, including prefix caching, KV‑cache configuration, attention backends, batching and quantization (FP8 / FP4 / NVFP4).

Optimize for NVIDIA GPUs and edge platforms (Blackwell‑based, e.g., Jetson Thor), working within memory‑bandwidth and hardware‑feature constraints.

Address latency, memory, and compute constraints for real‑time inference on edge devices

Apply knowledge distillation, pruning, and quantization techniques to reduce model size while maintaining performance

Run load, concurrency and accuracy regression tests every release cycle, and document what worked and what did not.

Collaborate with hardware and platform teams to ensure seamless integration of AI models into edge products

Required Skills & Qualifications -

Core technical skills (must‑have) -

LLM fine‑tuning: Hands‑on experience with SFT and parameter‑efficient methods (LoRA / QLoRA), plus preference optimization (e.g., DPO) where relevant. You should have taken fine‑tuned models from experiment to production.

ML stack: Strong Python, PyTorch and the Hugging Face ecosystem (Transformers, PEFT, TRL, Accelerate).

LLM inference and serving: Practical experience with vLLM, TensorRT-LLM or similar. This includes prefix caching, KV‑cache configuration, batching, attention backends and quantization (FP8 / FP4 / NVFP4).

GPU architecture and profiling: CUDA fundamentals, Nsight or similar profilers, and the ability to diagnose memory, bandwidth and compute bottlenecks.

Performance benchmarking: Ability to run load, concurrency and accuracy regression tests. You should measure TTFT, end‑to‑end latency and throughput, and weigh the accuracy‑versus‑latency trade‑off.

Evaluation design: Building benchmarks and golden datasets for LLMs and agents. Metrics include function‑calling accuracy, retrieval precision and task success rate.

Data pipelines: Data curation, transformation, synthetic data generation, and train / validation / test splits.

Deployment and infrastructure: Docker, Kubernetes and Linux‑based deployments, with familiarity with cloud or on‑prem GPU environments.

Customer‑facing and delivery skills (must‑have) -

Able to present benchmark results and recommendations to technical and business stakeholders, and to translate customer outcomes into model and infrastructure requirements.

Able to write technical documentation and operational guidance for handover, and to turn configurations and learnings into reusable accelerators.

Able to work with architects and other senior technical leads on design and best practices.

Able to coordinate with SME’s on roadmap, support and issue escalation.

Strong communication skills and comfort working embedded with customers.

Nice‑to‑have -

Direct experience with NVIDIA Nemotron, NeMo, NIM or NeMo Guardrails.

Experience with edge NVIDIA platforms (Jetson‑class) and their constraints, such as memory bandwidth and missing hardware features.

RAG, agentic frameworks (LangChain, LlamaIndex, AutoGen) and tool / function calling.

NL‑to‑SQL or small language model (SLM) evaluation.

Distributed training (DeepSpeed, FSDP), custom kernels, and experiment tracking (MLflow, W&B).

Async and non‑blocking service design (e.g., FastAPI, asyncio) for LLM applications.Open‑

source contributions, publications or talks.

Education & Experience -

Required: Bachelor's degree in Computer Science, Electrical/Electronics Engineering, Data Science, AI/ML, Mathematics, Statistics or a related quantitative field.