TECHNICAL PRODUCT LEAD

Back to Jobs

Job Description

TECHNICAL PRODUCT LEAD
Remote
Full-time

Patent Domain Requirements

What you must know cold
 Claim structure
 Claim charts
 Prior art
 Office actions
 Invalidity contentions
 Portfolio strategy
 Patent systems
 Claim structure: independent vs dependent claims, claim scope, means-plus-function limitations, preamble interpretation, claim differentiation doctrine
 Claim charts: limitation-by-limitation mapping for infringement and invalidity; understanding of reads on analysis; EOUs, IOUs, and claim construction approaches (plain meaning vs PHOSITA vs prosecution history)
 Prior art: §102 anticipation analysis, §103 obviousness (motivation to combine, PHOSITA level, secondary considerations), temporal relevance relative to priority date
 Office actions: reading and categorizing examiner rejections (§101, §102, §103, §112); claim amendment strategy; prosecution history and its estoppel implications; file wrapper analysis
 Invalidity contentions: structure and strategy of invalidity contentions in district court litigation; anticipation vs obviousness charts; motivation to combine arguments; §112 written description and enablement
 Portfolio strategy: forward/backward citation analysis; continuation strategy; patent family grouping; FTO analysis methodology; portfolio pruning criteria
 Patent systems: USPTO PAIR/Patent Center; EPO OPS; WIPO PCT; CPC and IPC classification systems; Derwent/Orbit for prior art search; global filing procedures

Credentials that signal domain depth
 USPTO registered patent agent (registration number required) OR JD from accredited law school with substantial patent prosecution or litigation experience
 Minimum 3 years hands-on experience creating claim charts, invalidity contentions, freedom-to- operate analyses, or office action responses in a professional context
 Demonstrable familiarity with at least one patent analytics platform (Patlytics, Derwent Innovation, Orbit Intelligence, PatSnap, Google Patents Advanced)
 Technical undergraduate degree preferred (engineering, chemistry, computer science, biology) —the ability to read a technical patent in semiconductors, software, biotech, or mechanical engineering is required
 Experience working directly with patent attorneys as a client, colleague, or counterpart —understanding of how Am Law firms and in-house IP teams work is essential

AI & LLM Requirements
What you must know cold
 LLM architecture: context windows, token limits, system/user/assistant patterns, temperature, sampling — and how each affects patent analysis output quality and consistency
 Prompt engineering: chain-of-thought prompting, few-shot examples with patent claim examples, structured JSON output schemas, negative constraints to prevent hallucination in legal contexts
 Tool use / function calling: designing tool definitions for patent database queries, prior art retrieval, citation lookup, USPTO PAIR API calls — the core of agentic patent workflows
 Multi-step agents: LangGraph or equivalent for stateful pipelines where prior art search feeds into claim chart generation feeds into citation verification — you can specify this architecture, not just describe it vaguely
 Evaluation: LLM-as-judge patterns, golden dataset design, eval metrics for legal accuracy (citation correctness, claim scope fidelity, argument soundness), Braintrust or equivalent eval tooling
 Hallucination in legal contexts: citation fabrication, claim scope inflation, temporal relevance errors in prior art, jurisdiction-specific legal standard drift — and how to mitigate each in prompt design
 Cost and latency: Anthropic prompt caching, token attribution per workflow, batching strategies for high-volume claim charting — you understand the economics of LLM features at enterprise scale

AI tools you will use daily
 Anthropic Claude API — primary LLM for all patent workflows; you understand model strengths, rate limits, caching, and the difference between Claude models
 LangSmith — for reviewing eval run results, tracking prompt quality over time, and making go/no-go decisions based on eval data
 LangGraph or equivalent — for specifying multi-step agent architectures to the engineering team (you spec, they implement)
 GitHub, JIRA and other productivity tools — async-first delivery management, spec writing, stakeholder communication

Product Leadership Requirements
 5+ years in B2B SaaS product management, with at least 2 years at a company competing in a domain with significant legal or regulatory accuracy requirements
 Proven track record shipping AI-powered features to enterprise legal or IP clients (Am Law firms, Fortune 500 in-house teams, or equivalent)
 Experience running a product without a Scrum Master or dedicated QA team — async-first, output- oriented delivery with direct engineering accountability
 Demonstrated ability to write LLM workflow specifications precise enough for engineers to implement without clarification sessions
 Strong enterprise client relationship skills: you can demo a half-built product, field hard questions from a GC or IP litigation partner, and translate that feedback into a specific prompt change
 Competitive analysis fluency: you actively use generally known products in this space regularly —you know where their claim charts fail and where their UX creates friction
 Ability to make technical tradeoff decisions alongside engineering: latency vs accuracy, prompt caching vs freshness, context window vs cost — without deferring to engineering to decide for you
 Track record of prioritizing with extreme discipline in resource-constrained environments — you understand that 10 engineers competing with a 100-person org means every feature must be 10×the value, not 10× the scope

Job Information

Rate / Salary

$ - $

Sector

Not Specified