{"id":3259319,"url":"https://alion.io/job/didi-global-aisuanfaanquanzhuanjia","title":"AI算法安全专家","company":{"id":231,"name":"DiDi Global","domain":"didiglobal.com","url":"https://alion.io/company/didi-global","size_band":"5000+","is_staffing_agency":false,"employer_type":"direct","is_intermediary":false,"listed_via":null,"ats_vendor":"Career site","truth_index":{"grade":"A","score":98,"open_postings":82,"ghost_share":0,"stale_share":0.085,"repost_share":0.012,"time_to_fill_p50_days":26,"computed_at":"2026-11-06T05:58:05Z"}},"role":"AI/ML","role_family":"AI/ML","seniority":null,"employment_type":null,"work_mode":"on_site","remote_scope":null,"remote_scope_basis":null,"remote_working_hours":null,"hiring_geo_confidence":"structured","locations":["Hangzhou, China"],"countries":["CN"],"hiring_countries":[],"hiring_countries_total":0,"salary":null,"salary_estimate":null,"experience_years_min":null,"visa_sponsorship":false,"relocation_package":false,"has_equity":false,"technologies":[{"name":"DPO","optional":false},{"name":"Java","optional":false},{"name":"Mixture of Experts","optional":false},{"name":"RLHF","optional":false},{"name":"SFT","optional":false}],"status":"live","first_seen_at":"2026-09-18T22:30:27Z","employer_posted_date":"2026-10-26","last_verified_at":"2026-11-06T05:56:43Z","board_verified":true,"closed_at":null,"days_open":48,"trust":{"level":"ok","repost_count":null,"flags":[],"days_open":48},"description":"岗位职责 / Responsibilities\n1.主导金融场景下大模型安全对齐的工程化方案设计与迭代。深入优化RLHF/DPO在垂域模型中的训练效率与对齐效果，构建覆盖输入过滤、意图识别、输出审查的多层级护栏系统，确保模型拒答率与误伤率在千万级日调用下达到金融级平衡点。\n2.跟进多模态大模型（图文/音视频）在远程面审、双录质检场景的落地节奏，提前储备并落地低成本、低延迟的多模态安全检测能力。研究方向包括Deepfake检测、跨模态一致性校验、视觉对抗样本防御等，确保多模态攻击面在金融场景中得到有效管控。\n3.建立AI安全内部评测标准体系。设计覆盖\"诈骗诱导、隐私泄露、合规红线、业务逻辑绕行\"等金融特有维度的安全测试集，并推动评测流程嵌入模型发布门禁\n4.主导构建面向金融黑灰产的自动化红队评测平台。不仅要自己编写越狱及提示注入的自动化攻击脚本，更要设计一套持续对抗训练Pipeline，让防御模型在与红队的\"左右互搏\"中自动进化，有效抵御针对金融场景的定制化诱导攻击\n5.攻克大模型安全领域在工程化中的痛点，如大模型推理下安全检测的毫秒级延迟要求、护栏系统在高并发场景下的稳定性等。\n任职要求 / Requirements\n1.深入理解大模型原理（Transformer、MoE、SFT、RLHF/DPO），有实际垂域模型（>10B参数）微调与对齐经验，能独立解决训练中的Loss震荡、模型崩溃等问题; 对大模型安全对齐、越狱攻防、内容安全、Agent安全中的至少两个方向有系统性实战经验。\n2. 对LLM攻击手法有系统性的实战认知。能构造绕过主流安全护栏的Prompt，深刻理解黑灰产在金融场景下的攻击动机与链路，具备以攻击者视角设计防御的思维习惯。\n3. 熟练使用Python/Java，熟悉高性能服务框架，能独立将PyTorch模型封装为高并发、低延迟的在线服务，并有实际压测与性能调优经验; 熟悉分布式训练框架，了解模型量化、蒸馏等推理优化技术。\n4. 有设计复杂安全系统的经验，能在高并发、强一致性要求的金融环境中，权衡\"安全强度\"与\"用户体验\"，拿出切实可行的落地方案（而非纯学术方案）。\n5. 善于利用AI智能体将团队研发效能提升30%以上，并有推动团队AI化转型的实际经验。\n6. 能与算法平台组、SRE、业务风控团队高效协同，将安全需求转化为通用的平台能力。","description_format":"text","description_chars":993,"description_truncated":false,"requirements":{"experience_years_min":null,"management_years_min":null,"team_size_min":null,"manages_managers":false,"education":null,"security_clearance":false,"languages":[{"language":"Chinese","level":"Advanced (C1)","optional":false}]},"benefits":[],"hiring_locations":[],"hiring_excludes":[],"relocation_offered":false,"industries":["Shared Mobility","Ride Hailing"],"lifecycle":[{"event":"open","at":"2026-10-26T03:32:16Z"}],"visa":[],"liveness":{"score":18,"band":"cold","label":"Long shot","p_open":1,"p_active":0.499,"p_room":0.36,"age_days":48,"expected_fill_days":26,"reasons":["conf:0","velocity","win:tail","crowd:brand"],"computed_at":"2026-11-06T05:58:05Z"},"pay":null,"html_url":"https://alion.io/job/didi-global-aisuanfaanquanzhuanjia","json_url":"https://alion.io/job/didi-global-aisuanfaanquanzhuanjia.json","meta":{"generated_at":"2026-11-06T06:14:32Z","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":10953,"day_limit":null,"remaining_today":null,"minute_limit":300,"resets_at":"2026-11-07T00:00:00Z"}},"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":231},"rest":"https://alion.io/mcp/rest/get_company?id=231"},"human":"https://alion.io/catalog?offer=company.slices&for=job%2Fdidi-global-aisuanfaanquanzhuanjia"},{"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%2Fdidi-global-aisuanfaanquanzhuanjia"},{"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%2Fdidi-global-aisuanfaanquanzhuanjia"},{"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/didi-global-aisuanfaanquanzhuanjia\", \"note\": \"what you need it for\"}"},"human":"https://alion.io/catalog?offer=company.verify&for=job%2Fdidi-global-aisuanfaanquanzhuanjia"}],"agent":{"mcp":"https://alion.io/mcp/jobs","mcp_open":"https://alion.io/mcp/open","next":[{"tool":"get_job","arguments":{"id":3259319}},{"tool":"search_jobs","arguments":{"company":"didiglobal.com","detail":"compact"}},{"tool":"match_people","arguments":{"job_id":3259319},"note":"people open to offers who fit this vacancy"}],"note":"The same data through MCP tools: https://alion.io/mcp/open needs no sign-in (compact searches, postings, people); https://alion.io/mcp with an Alion account has everything. Docs: https://alion.io/developers"}}