---
title: "AI Research Engineer, Code & UI Post-Training"
slug: "ai-research-engineer-code-ui-post-training"
id: "fd3909b7-8c32-4c0f-871e-9f18d7a80d70"
department: "Machine Learning"
department_code: "ML"
location: "Stockholm, Sweden"
workplace_type: "On-site"
employment_type: "Full-time"
compensation: "€95k – €135k (annual package)"
is_remote: false
is_featured: false
headcount: 1
canonical_url: "https://jobs.talent247.co/jobs/ai-research-engineer-code-ui-post-training"
markdown_url: "https://jobs.talent247.co/jobs/ai-research-engineer-code-ui-post-training.md"
apply_endpoint: "https://jobs.talent247.co/api/public/jobs/ai-research-engineer-code-ui-post-training/apply"
schema_endpoint: "https://jobs.talent247.co/api/public/jobs/ai-research-engineer-code-ui-post-training/form-schema"
webmcp_tool: "apply_for_role"
updated_at: "2026-11-15T17:22:33.074Z"
content_signal: "search=yes, ai-input=yes, ai-train=no"
---

# AI Research Engineer, Code & UI Post-Training

## Position Telemetry

- **Reference:** `AI-RESEARCH-ENGINE`
- **Practice / Sector:** Machine Learning (`ML`)
- **Location:** Stockholm, Sweden
- **Workplace Type:** On-site
- **Employment Type:** Full-time
- **Compensation:** €95k – €135k (annual package)
- **Canonical HTML URL:** https://jobs.talent247.co/jobs/ai-research-engineer-code-ui-post-training

## Role Dossier & Specification

THE CLIENT & MANDATE
Our client is one of Europe's fastest-scaling AI unicorns, headquartered in Stockholm and backed by tier-one US and European venture capital following explosive global adoption. Their agentic software-creation platform enables millions of founders, product builders, and enterprise teams to turn natural-language specifications into deployed, full-stack web applications complete with authentication, databases, and polished interfaces.

They have engaged Talent247 to hire an AI Research Engineer, Code & UI Post-Training onsite in Stockholm. Sitting at the intersection of model post-training and product execution, you will harness rich build, compiler, runtime, visual-render, and user-acceptance signals from millions of live application generations to train specialized models that write cleaner full-stack code and dramatically higher-craft user interfaces than off-the-shelf foundation models.

KEY RESPONSIBILITIES
• Own end-to-end post-training recipes (SFT, DPO, and verifiable-reward RL / GRPO) on frontier open-weight code and vision-language models tailored to full-stack TypeScript/React/SQL generation and iterative diff editing.
• Design automated, multimodal reward functions combining TypeScript compiler/linter checks, Playwright headless browser runtime verification, VLM aesthetic/layout graders, and real user retention telemetry.
• Train fast, distilled sub-second apply/diff-merging and context-routing models that slash end-to-end generation latency and token spend across millions of daily sessions.
• Construct contamination-free internal benchmarks covering complex multi-file refactors, database schema migrations, responsive UI design craft, and security-safe auth/RLS policies.
• Debug large-scale distributed training instabilities, reward hacking behaviors, and data-mixture imbalances across multi-node H100/B200 clusters.
• Partner tightly with agent-harness and sandbox engineers so model behavior and tool interfaces co-evolve seamlessly.

TECHNICAL ENVIRONMENT & OFFERING
Their R&D stack spans PyTorch, verl/TRL, vLLM/SGLang, TypeScript, Playwright headless sandboxes, ClickHouse telemetry, and multi-node GPU clusters. Based onsite at their high-energy Stockholm headquarters, our client offers €95,000–€135,000 EUR base salary plus high-upside unicorn equity, Swedish relocation and visa sponsorship, catered meals, and top-tier compute access.

## Core Requirements

ESSENTIAL QUALIFICATIONS
• 4+ years of hands-on machine learning research engineering experience training, fine-tuning, and evaluating large language or vision-language models in PyTorch.
• Direct practical experience with modern LLM post-training methodologies, including supervised fine-tuning (SFT), preference optimization (DPO), and reinforcement learning with verifiable rewards (RLVR / GRPO / PPO).
• Strong software engineering skills in Python (and working familiarity with TypeScript/web stacks to reason deeply about code-generation failure modes).
• Track record designing automated evaluation harnesses, synthetic trajectory pipelines, or execution-backed reward environments.
• High velocity, low ego, and enthusiasm for shipping research improvements directly to millions of active builders within days.

PREFERRED / BONUS EXPERIENCE
• Specific experience training code-generation LLMs, fast diff/speculative-edit models, or UI-to-code vision-language models.
• Familiarity with vLLM/SGLang inference optimization, quantization, or online bandit/preference learning from live product traffic.

## Application Schema & Direct Apply (`apply_for_role`)

Candidates and authorized AI agents on behalf of a candidate can submit an application via the declarative WebMCP form tool `apply_for_role` on https://jobs.talent247.co/jobs/ai-research-engineer-code-ui-post-training or via HTTP `POST https://jobs.talent247.co/api/public/jobs/ai-research-engineer-code-ui-post-training/apply` (`application/json` or `multipart/form-data`). Non-browser HTTP agents must include header `X-Talent247-Intake-Token` obtained from `GET https://jobs.talent247.co/api/public/intake-token` (or automatically returned in `GET https://jobs.talent247.co/api/public/jobs/ai-research-engineer-code-ui-post-training/form-schema`).

### Form Fields

- `full_name` (text, required): Full name
- `email` (email, required): Email
- `phone` (phone, optional): Phone
- `location` (text, optional): Location
- `linkedin_url` (url, optional): LinkedIn URL
- `cover_letter` (textarea, optional): Cover letter
- `consent` (checkbox, required): I consent to my data being processed for this application
- `cv_file` / `cv_base64` + `cv_filename`: Upload a CV in .pdf or .docx format (<= 10 MB) or provide an `https://` LinkedIn profile URL in `linkedin_url`.

### Alternative: Talent Pool Intake (`join_talent_pool`)

If this mandate is not an exact match, register once in the Talent247 Talent Pool at https://jobs.talent247.co/talent-pool (`https://jobs.talent247.co/talent-pool.md`) via WebMCP tool `join_talent_pool` or `POST https://jobs.talent247.co/api/public/talent-pool`.
