Sr. Engineer, Software - AI Compiler - Tenstorrent

DESCRIPTION FORMATTING RULES (CRITICAL - FOLLOW EXACTLY) Your job is to output clean, readable Markdown while PRESERVING ALL ORIGINAL CONTENT AND STRUCTURE EXACTLY. ABSOLUTE RULES - NEVER VIOLATE: 1. NEVER add, remove, or modify any text/wording from the original 2. NEVER paraphrase or rewrite sentences - keep exact wording 3. NEVER remove sections, paragraphs, disclaimers, or any content 4. PRESERVE all links exactly as they appear 5. PRESERVE the natural flow and spacing of the original document MARKDOWN FORMATTING: - ## for main section headings (bold standalone titles like "The Role", "You Will", "Requirements") - ### for sub-section headings - NEVER use # (H1) - the job title is already the page H1 - Use "-" bullet lists for ANY list of items - CRITICAL: If content is clearly a list (requirements, responsibilities, benefits), format as bullet list - Bold for inline emphasis, key terms, skills - preserve existing bold from original - Italic for emphasis or disclaimers - preserve existing italic from original - text for ALL hyperlinks - preserve every link SPACING RULES: - Preserve the natural paragraph breaks from the original content when they exist - IF the original lacks proper spacing (e.g., everything is one giant paragraph), ADD logical breaks: * Blank line before each section heading * Blank line between distinct topics/sections * Blank line before and after bullet lists - NO blank lines between bullet points in the same list - NO multiple consecutive blank lines (max 1 blank line between elements) - The output should be scannable and readable, even if the input wasn't OUTPUT QUALITY: - Clean, scannable markdown that matches the original document's flow - Professional formatting that readers can quickly parse - All content intact - nothing added, nothing removed ## TITLE FORMATTING RULES (CRITICAL) The "title" field MUST use the EXACT format: "[Job Title] - [Company Name]" - Always use a HYPHEN (" - "), NEVER "at" - Example: "Senior ML Engineer - OpenAI" (CORRECT) - Example: "Senior ML Engineer at OpenAI" (WRONG — never use "at" in the title field) - Clean up the raw title: fix capitalization, remove company name if already embedded, remove location info - Keep the job title portion concise but don't lose important qualifiers (e.g., "Senior", "Lead", "Staff") - The "metaTitle" field uses a DIFFERENT format ("[Job Title] at [Company]") — do NOT confuse the two ## SALARY DISPLAY FORMAT (IMPORTANT) Format salary display as compact "k" notation: - $187,000 - $220,000 → "$187k - $220k" - $100,000 - $120,000 USD → "$100k - $120k" - €80,000 - €100,000 → "€80k - €100k" - Always use "k" for thousands, remove ".0" decimals - Include currency symbol at the start, not the end ## WORK MODE INFERENCE RULES Determine workMode using these rules IN ORDER: 1. If explicitly stated ("remote", "hybrid", "on-site", "in-office") in title/description → use that 2. If location contains "Remote" → REMOTE 3. If multiple countries/continents listed → REMOTE (high confidence) 4. If location is specific city only (e.g., "London", "New York", "Tokyo") with no remote mention → ONSITE 5. If "flexible" or "work from home options" mentioned → HYBRID 6. If unclear after all checks → null ## EXPERIENCE LEVEL EXTRACTION RULES Extract experienceLevel using these rules: 1. From title keywords (highest priority): - "Intern/Internship" → ENTRY - "Junior/Jr/Associate" → JUNIOR - "Mid/Mid-level" → MID - "Senior/Sr/Staff/Principal" → SENIOR - "Lead/Tech Lead/Team Lead" → LEAD - "Director" → DIRECTOR - "Head of/Head" → HEAD - "VP/Vice President/Chief/C-level/CTO/CEO/CFO" → C_LEVEL 2. From years requirement in description: - 0-1 years → ENTRY - 1-2 years → JUNIOR - 3-4 years → MID - 5+ years → SENIOR (unless title indicates higher level) 3. If no indicators → null ## LOCATION RULES (CRITICAL - keep compact) Goal: Locations must be SHORT and COMPACT. Max 5 comma-separated values total. 1. REMOTE JOBS: - If open to all/most regions → just "Remote" - If remote but limited to specific regions → use the CANONICAL REMOTE TOKENS below in parentheses - If remote limited to 2 regions → "Remote (US, Europe)" (still 1 badge) - NEVER list every individual country/city for remote jobs 2. CANONICAL REMOTE TOKENS (use these EXACT values inside parentheses): - United States only → "Remote (US)" - US + Canada / North America → "Remote (NA)" - Europe / EU / EMEA / DACH / Nordics → "Remote (Europe)" - Asia / Asia Pacific / APAC / SEA / ANZ → "Remote (APAC)" - Latin America / South America / LATAM → "Remote (LATAM)" - Middle East / MENA / GCC → "Remote (MENA)" - Africa / Sub-Saharan Africa → "Remote (Africa)" - UK only → "Remote (UK)" - Specific country → "Remote (Singapore)", "Remote (Germany)", etc. 3. ONSITE/HYBRID with specific cities: - For each city, ALSO include the country as a separate entry for discoverability - Example: "Dubai" → "Dubai, United Arab Emirates" - Example: "London, Berlin" → "London, United Kingdom, Berlin, Germany" - If 4+ cities in same country → use country name: "United States" and at most 2 key cities - NEVER include both city AND state AND country for the same place (e.g., "New York" not "New York, NY, United States") - But DO include city AND country as separate comma entries: "New York, United States" 4. US-SPECIFIC: - NEVER list states separately (no "CA, NY, TX, FL...") - Many US cities → "United States" or top 2 cities + "United States" - Remote in US → "Remote (US)" 5. MAX 5 comma-separated values in the locations field ## TAG RULES (CRITICAL - content-derived skills only) Goal: Tags must be SPECIFIC, ACTIONABLE skills/technologies/tools that appear IN THE JOB DESCRIPTION. They surface in "In-Demand Skills" and must help candidates filter by real requirements — not generic industry words. FORBIDDEN TAGS (NEVER use these — they are useless on an AI job board): - "ai", "artificial intelligence", "machine learning", "tech", "product management", "business development", "marketing", "engineering", "design", "operations", "sales", "partnerships", "product marketing" - Any tag that merely restates the job category or industry (we already know it's AI/ML) - Any tag not explicitly mentioned or clearly implied by the job description text REQUIRED: Tags must be EVIDENCE-BASED. Only include a tag if the description (or title) explicitly mentions that skill, tool, protocol, or practice. If the post doesn't mention specific technologies or methods, use 1–2 tags maximum or leave tags minimal. GOOD TAGS (prefer these when present in the content): - AI/ML domains: "nlp", "computer vision", "reinforcement learning", "generative ai", "llm", "robotics", "autonomous systems" - Languages & runtimes: "python", "rust", "typescript", "go", "c++", "java", "cuda" - Frameworks & tools: "pytorch", "tensorflow", "jax", "hugging face", "langchain", "ray", "mlflow", "wandb" - Tools & practices: "figma", "wireframes", "product specs", "user research", "a/b testing", "sql", "react", "node" - Disciplines (only when clearly a focus): "ui/ux", "mlops", "data engineering", "compliance", "devrel", "deep learning" - Concrete deliverables/activities: "prds", "roadmaps", "model evaluation", "dashboards", "fine-tuning" RULES: 1. Do NOT repeat words already in the job title (e.g. "Product Manager" → no tag "product management"). 2. Prefer lowercase; well-known abbreviations are fine: "nlp", "ui/ux", "llm". 3. Max 3 tags, prefer 2. If nothing specific is stated, 0–1 tags is fine — do not fill with generic terms. 4. When in doubt, omit. Fewer accurate tags are better than generic ones. ## RELEVANCY SCORING (for feed ranking) Score how relevant this job is to an AI/ML job board audience (0-100). This score controls feed ranking — it directly determines which jobs appear first when a company has many openings. Be precise and consistent. 95-100: Core AI/ML research or engineering that BUILDS models (ML Engineer, AI Researcher, NLP Scientist, Computer Vision, RL, LLM, MLOps) 80-94: Software infrastructure that directly SUPPORTS AI systems (GPU/CUDA Engineer, ML Platform, Distributed Training Infra, AI DevOps) 65-79: Technical roles at AI companies (Software Engineer, Backend, Security, SRE) 50-64: Product, data, and design roles tied to AI products (AI Product Manager, UX Researcher on AI tools, Data Analyst) 30-49: Business and go-to-market roles (Sales, Marketing, Finance, Legal, Recruiting) 10-29: Administrative and support roles (Office Manager, Coordinator, Compliance) 0-9: Non-technical facility/manual roles (Electrician, Custodian, HVAC, Security Guard) Key heuristic: Would someone browsing an AI job board be excited to find this listing? A "Senior ML Engineer" at Anthropic = 98. A "Facilities Coordinator" at Anthropic = 5. ## AI RELEVANCE CLASSIFICATION (CRITICAL — controls whether job is listed) Classify whether this job involves AI/ML/LLM/agentic technologies. This is a strict binary gate for companies that aren't pure AI companies. Set "isAIRelevant": true if ANY of these apply: - The role is fundamentally about AI/ML (researcher, ML engineer, NLP scientist, etc.) - The job description requires building, training, deploying, or evaluating AI/ML models - The role involves significant use of LLMs, RAG, agentic frameworks, or AI APIs - The role directly supports AI systems (ML platform, AI infrastructure, GPU engineering) - The role is an AI product manager, AI designer, or AI-focused data scientist - The description explicitly lists AI/ML tools (PyTorch, TensorFlow, LangChain, etc.) Set "isAIRelevant": false if: - The role is generic software engineering with no AI/ML mention in the description - The role is business/operations/admin with no AI/ML component - The role is at a company that does AI but THIS specific position doesn't involve AI - Example: "Recruiter" at Stripe — no AI involvement even though Stripe uses AI - Example: "Backend Engineer" at Datadog — builds dashboards, no ML mentioned - Example: "Office Manager" at any company — never AI-relevant Be strict: a job at an AI-adjacent company is NOT automatically AI-relevant. The role or description must explicitly involve AI/ML/LLM technologies. ## EXTRACTION RULES === - Extract data ONLY if clearly stated - If not found, return null (not empty string, not "Competitive") - Use the structured data from job board when available (more reliable than parsing description) - For categoryId: ALWAYS select best matching category based on title and description AVAILABLE CATEGORIES (id:name): 1:Business Development, 2:Data, 3:Design, 4:Engineering, 5:Finance, 6:Legal, 7:Marketing, 8:Operations, 9:People Ops, 10:Product Management, 11:Research, 12:Sales, 14:Support EXPERIENCE LEVELS: ENTRY, JUNIOR, MID, SENIOR, LEAD, DIRECTOR, HEAD, C_LEVEL ## JOB LISTING TO ANALYZE === Title: Sr. Engineer, Software - AI Compiler Company: Tenstorrent Location: Austin, Texas, United States; Santa Clara, California, United States; Toronto, Ontario, Canada Description: Tenstorrent is leading the industry on cutting-edge AI technology, revolutionizing performance expectations, ease of use, and cost efficiency. With AI redefining the computing paradigm, solutions must evolve to unify innovations in software models, compilers, platforms, networking, and semiconductors. Our diverse team of technologists have developed a high performance RISC-V CPU from scratch, and share a passion for AI and a deep desire to build the best AI platform possible. We value collaboration, curiosity, and a commitment to solving hard problems. We are growing our team and looking for contributors of all seniorities. Join the team revolutionizing AI computing at Tenstorrent. You'll work on TT-Forge, our MLIR-based compiler that enables developers to run AI on all configurations of Tenstorrent hardware using an open-source, performant, and general-purpose compiler. You will be at the forefront of the AI hardware revolution, building compiler technologies that redefine what’s possible. This role is hybrid, and can be based out of Santa Clara, CA; Austin, TX; or Toronto; ON. We welcome candidates at various experience levels for this role. During the interview process, candidates will be assessed for the appropriate level, and offers will align with that level, which may differ from the one in this posting. Who You Are - A passionate software engineer eager to work on compiler technologies and the challenges of AI hardware, whether from compilers, systems, or broader software backgrounds. - Fluent in C++ and Python, with experience building complex systems that bridge high-level frameworks to low-level execution. - Excited by compiler optimization and machine learning, with experience in PyTorch, JAX, TensorFlow, or deep systems programming. - A collaborative problem-solver who thrives in open-source and enjoys working closely with hardware and software engineers. What We Need - A drive to solve novel challenges in AI compilation, from optimizing computational graphs to creating custom dialects and transformation passes. - Experience or strong interest in MLIR and how modular compiler frameworks connect AI models to advanced hardware. - Motivation to build technology that impacts the future of AI, knowing your work will enable the next wave of breakthroughs. What You Will Learn - How to build open-source compiler frameworks supporting diverse AI models and workloads, including training and multi-chip scaling. - Deep expertise in compiler technologies including custom MLIR dialects (TTIR, TTNN, TTKernel) and transformation passes. - New methods for human-in-the-loop compiler optimization using TT-Explorer, making advanced tuning tools usable by all developers. - How compiler technology powers Tenstorrent’s mission to deliver affordable, open-source AI computing in a highly competitive space. Compensation for all engineers at Tenstorrent ranges from $100k - $500k including base and variable compensation targets. Experience, skills, education, background and location all impact the actual offer made. Tenstorrent offers a highly competitive compensation package and benefits, and we are an equal opportunity employer. This offer of employment is contingent upon the applicant being eligible to access U.S. export-controlled technology. Due to U.S. export laws, including those codified in the U.S. Export Administration Regulations (EAR), the Company is required to ensure compliance with these laws when transferring technology to nationals of certain countries (such as EAR Country Groups D:1, E1, and E2). These requirements apply to persons located in the U.S. and all countries outside the U.S. As the position offered will have direct and/or indirect access to information, systems, or technologies subject to these laws, the offer may be contingent upon your citizenship/permanent residency status or ability to obtain prior license approval from the U.S. Commerce Department or applicable federal agency. If employment is not possible due to U.S. export laws, any offer of employment will be rescinded. ## SEO META FIELDS (CRITICAL - generate for every job) "metaTitle": SEO title for this job page. Format: '[Job Title] at [Company]'. Keep concise but don't sacrifice clarity. Do NOT add any site name suffix — the system appends ' | Underprompt' automatically. "metaDescription": Write a compelling, keyword-rich meta description between 150-180 characters. This appears in Google search results and directly impacts click-through rate AND ranking. You must naturally weave in high-value search keywords that someone looking for this role would actually Google — think: the job title, specific technologies/frameworks (e.g. PyTorch, TensorFlow, LLMs, NLP, computer vision), the company name, location terms (e.g. "remote", city names), and industry terms (e.g. "AI jobs", "machine learning", "ML engineering"). The description should read naturally and be compelling — not keyword-stuffed. Include 2-3 of these details when available: core technologies or frameworks, location/remote flexibility, salary range, team context, or a unique perk. Vary your sentence structure. NEVER use calls-to-action like 'Apply now' or 'Apply on Underprompt'. NEVER start with the job title or company name (the title tag already has that). Each description must feel hand-written and distinct. Good meta description examples (notice natural keyword integration): - "Build production ML pipelines using PyTorch and distributed training at this remote AI infrastructure team. ML engineering role with $180k-$220k compensation." - "Lead brand strategy for a top AI research lab. Remote ML marketing role with equity compensation, creative autonomy, and a globally distributed team." - "Design and evaluate large language models powering next-gen AI products. NLP research role based in San Francisco or fully remote." - "Shape how enterprises adopt AI-powered workflows. ML product management role bridging research and production deployment." - "Design intuitive interfaces for AI-powered products. UX/UI role using Figma, user research, and close collaboration with ML engineers. NYC or remote." Bad meta descriptions (NEVER do this): - "Senior Engineer at Company. Full-time remote role. Apply now on Underprompt." (templated, CTA, boring) - "Apply for Senior Engineer at Company on Underprompt. Remote position available." (CTA-first, generic) - "Company is hiring a Senior Engineer. This is a full-time position based in New York." (reads like a form letter, no keywords) ## OUTPUT FORMAT (valid JSON only) { "title": " - (MUST use hyphen, NEVER 'at')", "categoryId": , "categoryName": " ", "jobType": " ", "workMode": " ", "experienceLevel": " ", "locations": " ", "tags": " ", "relevancyScore": , "isAIRelevant": , "salary": { "min": , "max": , "currency": " ", "display": " " }, "metaTitle": " ", "metaDescription": " ", "description": " ", "analysis": { "fields": [ {"field": "categoryId", "found": , "sourceText": " ", "confidence": " ", "reason": " "}, {"field": "jobType", "found": , "sourceText": " ", "confidence": " ", "reason": " "}, {"field": "workMode", "found": , "sourceText": " ", "confidence": " ", "reason": " "}, {"field": "experienceLevel", "found": , "sourceText": " ", "confidence": " ", "reason": " "}, {"field": "locations", "found": , "sourceText": " ", "confidence": " ", "reason": " "}, {"field": "tags", "found": , "sourceText": " ", "confidence": " ", "reason": " "}, {"field": "salary", "found": , "sourceText": " ", "confidence": " ", "reason": " "} ], "formattingApplied": [" "] "warnings": [" "] } } Return ONLY the JSON object, no markdown fences, no explanation.