{"id":3097213,"url":"https://alion.io/job/recho-senior-llm-training-engineer-post-training","title":"Senior LLM Training Engineer (Post-training)","company":{"id":2675564,"name":"Recho","domain":"recho-ai.com","url":"https://alion.io/company/recho-ai","size_band":"51-200","is_staffing_agency":false,"employer_type":"direct","is_intermediary":false,"listed_via":null,"ats_vendor":"HERP","truth_index":{"grade":"A","score":87,"open_postings":13,"ghost_share":0,"stale_share":0.538,"repost_share":0,"time_to_fill_p50_days":null,"computed_at":"2026-10-24T06:02:35Z"}},"role":"AI/ML","role_family":"AI/ML","seniority":"senior","employment_type":null,"work_mode":"hybrid","remote_scope":null,"remote_scope_basis":null,"remote_working_hours":null,"hiring_geo_confidence":"structured","locations":["Chiyoda, Japan"],"countries":["JP"],"hiring_countries":[],"hiring_countries_total":0,"salary":{"min":10000000,"max":20000000,"currency":"JPY","period":"year","gross":null,"usd_annual":128560},"salary_estimate":null,"experience_years_min":null,"visa_sponsorship":true,"relocation_package":false,"has_equity":true,"technologies":[{"name":"A/B Testing","optional":false},{"name":"AI Agents","optional":false},{"name":"DPO","optional":false},{"name":"Function Calling","optional":false},{"name":"GRPO","optional":false},{"name":"Kubernetes","optional":false},{"name":"LLM","optional":false},{"name":"Post-training","optional":false},{"name":"PPO","optional":false},{"name":"Pre-training","optional":false},{"name":"Reinforcement Learning","optional":false},{"name":"Reward Modeling","optional":false},{"name":"RLHF","optional":false},{"name":"SFT","optional":false},{"name":"Synthetic Data","optional":false},{"name":"Claude Code","optional":true},{"name":"DeepSpeed","optional":true},{"name":"FSDP","optional":true},{"name":"KV Cache","optional":true},{"name":"Megatron-LM","optional":true},{"name":"Multimodal AI","optional":true},{"name":"OpenAI Codex","optional":true},{"name":"Python","optional":true},{"name":"PyTorch","optional":true},{"name":"Quantization","optional":true},{"name":"SGLang","optional":true},{"name":"Slack","optional":true},{"name":"SLURM","optional":true},{"name":"Speculative Decoding","optional":true},{"name":"Tokenization","optional":true},{"name":"Transformers","optional":true},{"name":"TRL","optional":true},{"name":"vLLM","optional":true},{"name":"Weights & Biases","optional":true}],"status":"live","first_seen_at":"2026-10-22T15:44:52Z","employer_posted_date":"2026-10-22","last_verified_at":"2026-10-24T07:25:54Z","board_verified":true,"closed_at":null,"days_open":1,"trust":{"level":"ok","repost_count":null,"flags":[],"days_open":1},"description":"Senior LLM Training Engineer (Post-training)\nDescription\n## In short\nYou'll make Recho's AI models smarter for our use cases through post-training. You'll design the training methods, the reward design that \"grades\" the model, and the training data yourself, and you'll own the model's intelligence itself. Your job is to optimize general-purpose LLMs for our use cases, and down the line, we plan to offer that performance externally as an API business.\n- **Team language:** English (Japanese not required)\n- **Salary:** ¥10M-¥20M per year, plus stock options\n- **Location:** Tokyo (Hanzomon), hybrid. Candidates based outside Japan are welcome. We offer visa sponsorship and relocation support, and we're open to discussing fully remote arrangements.\n## You might be a fit if you are…\n- An ML engineer or researcher who has worked deeply on LLM training, especially post-training\n- Someone who has applied SFT, preference optimization or RLHF to a production model and shipped it\n- Someone with a CS or math background (MS/PhD, or equivalent industry experience) and a track record of conference publications or open-source work\n## About the role\nAt Recho, we don't use general-purpose LLMs as they are. We optimize them for our use cases through post-training and build them deep into our product. Your core mission is to design, implement and run training methods such as SFT, preference optimization and reinforcement learning (RL), and to own training quality for post-training of a single model.\nYou'll work at the core of what makes the model smart. That includes reward design (reward models, RLVR, LLM-as-a-Judge), designing training data quality and mixture ratios, and improving the post-training algorithms themselves (loss functions, KL penalty, advantage estimation, training stability).\nWe also plan to offer the performance we gain here externally as an API and grow it into a second pillar of our business. The scope includes post-training for agentic use cases such as long CoT and reasoning, tool use and function calling.\n## What this role is *not*\n- Not an on-call or SRE role. SLA management, incident recovery and infrastructure maintenance for online services are not the focus.\n- GPU cluster operations and general SRE/DevOps are handled by a separate team.\n- Building the infrastructure that runs training and inference (rollout systems, weight sync, GPU allocation) belongs to our RL Infrastructure Engineer role. You'll own the algorithms, rewards and data, and partner with that team.\n- Pre-training from scratch is not the main job. The focus is post-training.\n## What you'll do\n### 1. Design, implement and run post-training\n- Design, implement and run SFT, preference optimization (DPO, KTO, ORPO, SimPO, etc.) and online RL (PPO, GRPO, RLOO, etc.)\n- Post-training for long CoT and reasoning, tool use, function calling and agentic use cases\n- Improve post-training algorithms: loss functions, KL penalty, advantage estimation, training stability and sample efficiency\n### 2. Design rewards\n- Build reward models, RLVR (verifiable rewards) and LLM-as-a-Judge pipelines\n- Detect, analyze and address reward hacking and reward over-optimization\n### 3. Manage training data quality\n- Generate synthetic data and manage data quality: mixture optimization, difficulty control, deduplication and benchmark contamination checks\n- Generate and filter preference and synthetic data\n### 4. Build evaluation in-house\n- Design new internal evaluation sets that correlate with product quality\n- Design and build our evaluation platform in-house, and design human evaluation, A/B tests and annotation guidelines\nWe're happy to go into more detail during the intro call and interviews.\n## Why this role\n- **Own the core of what makes the model smart.** You'll own the algorithms, rewards and data that optimize general-purpose LLMs for our use cases, and you'll be responsible for post-training quality. You won't just use a platform; you'll design the model's intelligence itself.\n- **Build a second business pillar from scratch.** We plan to offer the performance gained through post-training as an API and grow it into a second pillar of the business. Your work spans product optimization all the way to a new business.\n- **Run cutting-edge methods in production.** You'll apply top-lab methods, such as online RL like GRPO, long CoT and reasoning, and agentic post-training, directly to a product that major enterprises run in production.\n## Tech stack\n- **Languages & frameworks:** Python, PyTorch\n- **Training methods:** SFT, DPO, KTO, ORPO, SimPO, PPO, GRPO, RLOO\n- **Rewards & evaluation:** Reward models, RLVR (verifiable rewards), LLM-as-a-Judge\n- **RL & distributed training:** verl, OpenRLHF, TRL, FSDP, DeepSpeed, Megatron-LM\n- **Rollout inference:** vLLM, SGLang\n- **Experiment tracking & compute:** Weights & Biases, etc.; Slurm, Kubernetes\n## Our product\nWe build the Recho AI Voice Platform, which delivers voice conversations natural enough to be hard to tell apart from a human. We also deploy and operate Voice AI Agents for large enterprises.\nOur edge is a platform for building and improving high-quality agents quickly. It captures industry know-how and turns user feedback into a fast, continuous improvement loop. Our agents now handle complex, large-scale contact center operations in finance, manufacturing and the public sector. In some deployments, they've resolved every call without handing off to a human agent.\n## Hiring process\nIntro call → Case assignment & first interview → Paid work trial* → Final interview → Offer meeting\n*The work trial is 1-2 consecutive days at our Tokyo office, working on real tasks alongside the team, and it's paid. It's a chance for both sides to check fit and get a feel for the work.\n- **If you're based outside Japan,** we'll agree on the format together. We won't ask you to travel to Japan just for this step.\n- If your schedule or your current employer's side-work policy makes the trial difficult, just let us know.\n- We may also ask for references.\n## About Recho\nRecho develops foundation models for customer-support AI agents and also deploys them for clients, all within one organization. Our core strength is voice AI technology we built entirely in-house, which resolves the industry's long-standing trade-off between quality and cost.\nWe've closed a Series A of approximately ¥700M and are rapidly rolling out to demanding production environments at major companies such as SBI Shinsei Bank and KDDI.\n## Learn more\n- [Corporate website](https://recho-ai.com/)\n- [Team interviews (note, in Japanese)](https://note.com/recho_ai)\n- [Company deck](https://recruit.recho-ai.com/)\n- [Case study: Partnership with KDDI and Altius Link (Japanese)](https://prtimes.jp/main/html/rd/p/000000021.000159800.html)\n- [Case study: SBI Shinsei Bank (Japanese)](https://mainichi.jp/articles/20260130/k00/00m/020/275000c)\n- [2-minute voice demo](https://recruit.recho-ai.com/#2f276fb799f480d5bba7d5784d6ad31e) - give it a listen!\nRequirements\n## In short\nYou'll make Recho's AI models smarter for our use cases through post-training. You'll design the training methods, the reward design that \"grades\" the model, and the training data yourself, and you'll own the model's intelligence itself. Your job is to optimize general-purpose LLMs for our use cases, and down the line, we plan to offer that performance externally as an API business.\n- **Team language:** English (Japanese not required)\n- **Salary:** ¥10M-¥20M per year, plus stock options\n- **Location:** Tokyo (Hanzomon), hybrid. Candidates based outside Japan are welcome. We offer visa sponsorship and relocation support, and we're open to discussing fully remote arrangements.\n## You might be a fit if you are…\n- An ML engineer or researcher who has worked deeply on LLM training, especially post-training\n- Someone who has applied SFT, preference optimization or RLHF to a production model and shipped it\n- Someone with a CS or math background (MS/PhD, or equivalent industry experience) and a track record of conference publications or open-source work\n## About the role\nAt Recho, we don't use general-purpose LLMs as they are. We optimize them for our use cases through post-training and build them deep into our product. Your core mission is to design, implement and run training methods such as SFT, preference optimization and reinforcement learning (RL), and to own training quality for post-training of a single model.\nYou'll work at the core of what makes the model smart. That includes reward design (reward models, RLVR, LLM-as-a-Judge), designing training data quality and mixture ratios, and improving the post-training algorithms themselves (loss functions, KL penalty, advantage estimation, training stability).\nWe also plan to offer the performance we gain here externally as an API and grow it into a second pillar of our business. The scope includes post-training for agentic use cases such as long CoT and reasoning, tool use and function calling.\n## What this role is *not*\n- Not an on-call or SRE role. SLA management, incident recovery and infrastructure maintenance for online services are not the focus.\n- GPU cluster operations and general SRE/DevOps are handled by a separate team.\n- Building the infrastructure that runs training and inference (rollout systems, weight sync, GPU allocation) belongs to our RL Infrastructure Engineer role. You'll own the algorithms, rewards and data, and partner with that team.\n- Pre-training from scratch is not the main job. The focus is post-training.\n## What you'll do\n### 1. Design, implement and run post-training\n- Design, implement and run SFT, preference optimization (DPO, KTO, ORPO, SimPO, etc.) and online RL (PPO, GRPO, RLOO, etc.)\n- Post-training for long CoT and reasoning, tool use, function calling and agentic use cases\n- Improve post-training algorithms: loss functions, KL penalty, advantage estimation, training stability and sample efficiency\n### 2. Design rewards\n- Build reward models, RLVR (verifiable rewards) and LLM-as-a-Judge pipelines\n- Detect, analyze and address reward hacking and reward over-optimization\n### 3. Manage training data quality\n- Generate synthetic data and manage data quality: mixture optimization, difficulty control, deduplication and benchmark contamination checks\n- Generate and filter preference and synthetic data\n### 4. Build evaluation in-house\n- Design new internal evaluation sets that correlate with product quality\n- Design and build our evaluation platform in-house, and design human evaluation, A/B tests and annotation guidelines\nWe're happy to go into more detail during the intro call and interviews.\n## Why this role\n- **Own the core of what makes the model smart.** You'll own the algorithms, rewards and data that optimize general-purpose LLMs for our use cases, and you'll be responsible for post-training quality. You won't just use a platform; you'll design the model's intelligence itself.\n- **Build a second business pillar from scratch.** We plan to offer the performance gained through post-training as an API and grow it into a second pillar of the business. Your work spans product optimization all the way to a new business.\n- **Run cutting-edge methods in production.** You'll apply top-lab methods, such as online RL like GRPO, long CoT and reasoning, and agentic post-training, directly to a product that major enterprises run in production.\n## Tech stack\n- **Languages & frameworks:** Python, PyTorch\n- **Training methods:** SFT, DPO, KTO, ORPO, SimPO, PPO, GRPO, RLOO\n- **Rewards & evaluation:** Reward models, RLVR (verifiable rewards), LLM-as-a-Judge\n- **RL & distributed training:** verl, OpenRLHF, TRL, FSDP, DeepSpeed, Megatron-LM\n- **Rollout inference:** vLLM, SGLang\n- **Experiment tracking & compute:** Weights & Biases, etc.; Slurm, Kubernetes\n## Our product\nWe build the Recho AI Voice Platform, which delivers voice conversations natural enough to be hard to tell apart from a human. We also deploy and o...","description_format":"text","description_chars":18916,"description_truncated":true,"requirements":{"experience_years_min":null,"management_years_min":null,"team_size_min":null,"manages_managers":false,"education":{"level":"phd","optional":true},"security_clearance":false,"languages":[{"language":"Japanese","level":"All levels","optional":false},{"language":"Chinese","level":"All levels","optional":true},{"language":"English","level":"All levels","optional":false}]},"benefits":["Annual leave","Apple Macbook","Stock options"],"hiring_locations":[{"name":"Japan","iso":"JP","kind":"country"}],"hiring_excludes":[],"relocation_offered":true,"industries":["Speech & Audio AI","AI Agents","Call Center Services"],"lifecycle":[{"event":"open","at":"2026-10-22T16:29:12Z"}],"visa":[],"liveness":{"score":63,"band":"ok","label":"Likely open","p_open":1,"p_active":0.632,"p_room":1,"age_days":1,"expected_fill_days":36,"reasons":["conf:2","stale_co","velocity","win:early"],"computed_at":"2026-10-24T06:02:35Z"},"pay":{"stated_usd_annual":128560,"is_top_pay":false},"html_url":"https://alion.io/job/recho-senior-llm-training-engineer-post-training","json_url":"https://alion.io/job/recho-senior-llm-training-engineer-post-training.json","meta":{"generated_at":"2026-10-24T11:19:28Z","cache_seconds":300,"methodology":"https://alion.io/methodology","terms":"https://alion.io/terms","contact":"https://alion.io/contact","api":"https://alion.io/developers","about":"Alion is a live layer of people, companies and AI agents: who they are, whether they are real and active right now, what they do and how to work with them, readable by people and by agents and paid per call.","catalog":"https://alion.io/catalog.json","usage":{"tier":"crawler_verified","counted_by":"address","units_charged":1,"used_today":16058,"day_limit":null,"remaining_today":null,"minute_limit":300,"resets_at":"2026-10-25T00:00:00Z"}},"offers":[{"id":"company.slices","title":"One company in depth, by slice","status":"live","price":{"credits":0.02,"usd":0.002,"plus_per_slice":{"credits":0.05,"usd":0.005}},"unit":"per company, plus each slice with data","note":"the employer in depth","call":{"mcp_tool":"get_company","arguments":{"id":2675564},"rest":"https://alion.io/mcp/rest/get_company?id=2675564"},"human":"https://alion.io/catalog?offer=company.slices&for=job%2Frecho-senior-llm-training-engineer-post-training"},{"id":"market.stats","title":"A market slice: pay, demand and time to fill","status":"live","price":{"credits":1,"usd":0.1},"unit":"per slice","note":"pay, demand and time to fill for this role and place","call":{"mcp_tool":"market_stats"},"human":"https://alion.io/catalog?offer=market.stats&for=job%2Frecho-senior-llm-training-engineer-post-training"},{"id":"job.search","title":"Open jobs by role, technology, place, pay and visa","status":"live","price":{"credits":0.02,"usd":0.002},"unit":"per posting in a list","note":"similar open postings","call":{"mcp_tool":"search_jobs"},"human":"https://alion.io/catalog?offer=job.search&for=job%2Frecho-senior-llm-training-engineer-post-training"},{"id":"company.verify","title":"Is this company real and active right now","status":"pilot","price":null,"unit":"per company","request":{"url":"https://alion.io/catalog/request","method":"POST","body":"{\"offer\": \"company.verify\", \"for\": \"job/recho-senior-llm-training-engineer-post-training\", \"note\": \"what you need it for\"}"},"human":"https://alion.io/catalog?offer=company.verify&for=job%2Frecho-senior-llm-training-engineer-post-training"}]}