Jabil
engineering, supply chain, manufacturing
Senior/StaffSLM&VLMEngineer—Post-Training,ToolCalling&Agents
Neural analysis suggests this role is
optimal for Senior candidates.
“Senior / Staff SLM & VLM Engineer — Post-Training, Tool Calling & Agents at Jabil. Skills: Small Language Models (SLMs), Vision-Language Models (VLMs), Post-Training, Tool Calling, Agents, Continuous Pretraining, Supervised Instruction Tuning (SFT), Compression, Distillation, Edge / Low-Latency Inference Optimization. lead the R&D of Small Language Models (SLMs) and Vision-Language Models (VLMs) for edge / low-latency and cost-efficient production scenarios. own the continuous pretraining, super”
What You'll Achieve.
deliver reliable, measurable improvements in inference efficiency, tool-use success rate, and overall model quality; improve task performance and domain adaptation; improving throughput and cost-per-token; improve success rate and ROI; align the model toward objectives including: semantic understanding, tool-use success rate, content generation quality and consistency; continuously improve training quality; improving both quality and efficiency over time
Industry & Context.
experimental discipline; failure analysis
Ability to communicate effectively in both Chinese (Mandarin) and English as the successful person will have to liaise with the our counterparts in China.
What They're Looking For.
Must Have
software engineering skills in Python and C ++, experience building ML training/evaluation pipelines in PyTorch, Hands-on experience in model efficiency and inference optimization (e. g. , distillation, quantization, pruning, serving optimization), Experience with high-performance computing and acceleration: CUDA and/or SIMD, profiling and performance tuning, Ability to read and reproduce key ideas from recent papers and implement algorithms with experimental discipline, Ability to communicate effectively in both Chinese (Mandarin) and English
What You'll Do.
lead the R&D of Small Language Models (SLMs) and Vision-Language Models (VLMs) for edge / low-latency and cost-efficient production scenarios
own the continuous pretraining
supervised instruction tuning (SFT)
and compression/distillation pipelines
work closely with platform teams to deliver reliable
measurable improvements in inference efficiency
tool-use success rate
and overall model quality
Conduct continuous pretraining and SFT for SLMs and VLMs to improve task performance and domain adaptation
Build reproducible training workflows in PyTorch
including data processing
Design and implement efficient compression strategies for SLM/VLM
including knowledge distillation
and quantization-oriented training or post-training optimization
Optimize model serving and inference for low-latency / edge scenarios by improving throughput and cost-per-token via techniques such as quantization
caching/KV optimizations
and decoding-time optimizations
Architect and implement a production-grade tool calling (function/tool calling) framework
Apply post-training methods such as PPO / DPO / GRPO-like optimization and reward modeling to align the model toward objectives
Support both offline and online iteration loops
including policy evaluation
and safe deployment gating
Design automated pipelines for data collection
labeling/weak supervision
and dataset version management to continuously improve training quality
Ensure datasets support both SFT and preference/RL style post-training
Build robust evaluation mechanisms: offline benchmarks
task suites for tool-use
and reliability metrics
Drive rapid iteration through A/B comparisons
improving both quality and efficiency over time
How You'll Work.
Team & Collaboration
work closely with platform teams to deliver reliable, measurable improvements; liaise with our counterparts in China
Communication Scope
Ability to communicate effectively in both Chinese (Mandarin) and English
Full Job Description
At Jabil (NYSE: JBL), we are proud to be a trusted partner for the world's top brands, offering comprehensive engineering, supply chain, and manufacturing solutions. With 60 years of experience across industries and a vast network of over 100 sites worldwide, Jabil combines global reach with local expertise to deliver both scalable and customized solutions. Our commitment extends beyond business success as we strive to build sustainable processes that minimize environmental impact and foster vibrant and diverse communities around the globe. **Job Summary** We are looking for a highly capable engineer/researcher to lead the R&D of **Small Language Models (SLMs)** and **Vision-Language Models (VLMs)** for **edge / low-latency** and cost-efficient production scenarios. You will own the **continuous pretraining, supervised instruction tuning (SFT)**, and **compression/distillation** pipelines, and work closely with platform teams to deliver reliable, measurable improvements in **inference efficiency, tool-use success rate, and overall model quality**. **Key Responsibilities** **1) SLM/VLM Training: Continuous Pretraining & Instruction Tuning (SFT)** * Conduct **continuous pretraining** and **SFT** for SLMs and VLMs to improve task performance and domain adaptation. * Build reproducible training workflows in **PyTorch** , including data processing, training, evaluation, and model versioning. **2) Compression, Distillation & Edge/Low-Latency Inference Optimization** * Design and implement **efficient compression** strategies for SLM/VLM, including **knowledge distillation** , pruning, and quantization-oriented training or post-training optimization. * Optimize model serving and inference for **low-latency / edge** scenarios by improving throughput and cost-per-token via techniques such as quantization, caching/KV optimizations, batching strategies, and decoding-time optimizations. **3) Tool Calling System: Catalog, Routing, Validation, Fallback & Observability** * Archite
Applying for this Senior / Staff SLM & VLM Engineer — Post-Training, Tool Calling & Agents role?
Most applicants get filtered before a human reads their resume. See if yours makes the cut.
How to Apply on Workday
- Workday has a multi-step form — save your progress after every section.
- "Apply With LinkedIn" can fail or lose data; manual entry is more reliable.
- Watch for the "Submit for Review" final step — hitting "Save" alone does not submit.
- Job requisition numbers are useful when following up with HR by email.
ANONYMOUS · UNFILTERED
What do employees actually say about Jabil?
Real rants from real employees. Read before you apply.