Remote AI Alignment Engineer (Part-Time, 7B–70B+)

DataJobs; Newport (RI); Remote; USD 138,000 - 344,000; Part time

Job summary

Wider Security LLC is seeking a part-time, fully remote AI Engineer to drive post-training and alignment for large language models at scale. You will lead end-to-end workflows from data prep to evaluation and deployment, focusing on 7B through 70B+ models in an async environment.

You will apply RLHF, DPO, and SFT with QLoRA/LoRA, and build safety evaluation pipelines while calibrating policy thresholds at inference time. Strong Python proficiency required.

Qualifications

• US citizenship and current US residency are required.
• Production experience with post-training open-weight LLMs at 7-8B, 13-30B, or 30B+ scales.
• Experience with open-weight families such as Llama, Qwen, Mistral or similar.
• Hands-on with QLoRA, LoRA, and efficient fine-tuning for large models.
• Background in alignment approaches including SFT, DPO, RLHF, RLAIF, or constitutional AI.
• Experience training models for reliable structured outputs (JSON/schema/output types).
• Familiar with serving stacks like vLLM, TGI, and logprob-level outputs; distributed training with DeepSpeed/FSDP/Megatron-LM.

Responsibilities

• Lead and contribute to post-training workflows including supervised fine-tuning, instruction tuning, DPO, RLHF, RLAIF, and related alignment techniques.
• Use QLoRA and other efficient fine-tuning methods across models from 7B to 70B+ parameters.
• Train models to produce reliable structured outputs under adversarial input conditions.
• Build and run evaluation pipelines for safety-critical behavior with adversarial test suites and red-team integration.
• Calibrate decision thresholds to support tiered policy configurations at the serving layer.
• Design training approaches that preserve inference-time policy specification without retraining.
• Curate and prepare training data, evaluation sets, and preference data pipelines.
• Iterate on training strategy to improve task performance, calibration, and adversarial robustness.
• Document approaches and decisions clearly for a distributed, async-first team.

Skills

Python

QLoRA

LoRA

RLHF

RLAIF

DPO

SFT

Garak

vLLM

TGI

DeepSpeed

Megatron-LM

Tools

Garak

logprob outputs

DeepSpeed

FSDP

Megatron-LM

Llama

Qwen

Mistral

Job description

Wider Security LLC is seeking a part-time, fully remote AI Engineer to drive post-training and alignment for large language models at scale. You will lead end-to-end workflows from data prep to evaluation and deployment, focusing on 7B through 70B+ models in an async environment.

You will apply RLHF, DPO, and SFT with QLoRA/LoRA, and build safety evaluation pipelines while calibrating policy thresholds at inference time. Strong Python proficiency required.