Full-Time

Senior AI Product Engineer

Code

Posted on 9/23/2025

Semgrep

Semgrep

201-500 employees

Static code vulnerability scanning for SDLC

Compensation Overview

$176k - $207k/yr

Boston, MA, USA + 3 more

More locations: San Francisco, CA, USA | New York, NY, USA | Denver, CO, USA

In Person

Category
Software Engineering (1)
Required Skills
LLM
Python
Machine Learning
Requirements
  • 5+ years of experience writing production software
  • Curiosity and a love of new technologies, especially AI/ML
  • Experience experimenting with GPT-4/GPT-5, Codex, Claude, or other LLMs; familiarity with ML algorithms and applied research
  • Comfort working in a fast-paced environment where prototypes are rapidly iterated or discarded
  • Strong Python skills (experience with other languages a plus)
  • Interest in prompt engineering, embeddings, and vector databases
  • Excitement about building for customers, iterating fast, and seeing solutions solve real developer problems
  • Excellent and proactive communication skills, both verbal and written
Responsibilities
  • Integrate AI platform APIs into the Code product
  • Develop and refine LLM prompt chains for real developer use cases
  • Experiment with the latest AI/ML advances and determine how they can be productized
  • Evaluate and fine-tune ML models using human- and machine-generated data
  • Learn directly from users to understand their needs and deliver features that help them secure their code
  • Work on major product initiatives end-to-end, from design and prototyping through implementation and deployment
  • Contribute to technical discussions, roadmap planning, and mentoring within the team
  • Building new detection capabilities that utilize fast, deterministic static analysis technologies within an autonomous LLM-driven agentic code analysis pipeline
  • Identifying causes of and solving for non-determinism in model output
  • Generating real-time, AI-powered remediation guidance for vulnerabilities in developers’ own code
  • Automatically drafting secure code suggestions that align with project context and frameworks
  • Building and orchestrating purpose-built custom agents for evaluating and triaging code security risks
  • Enhancing the IDE experience with AI-assisted security insights and explanations
  • Experimenting with AI-powered rule generation to help customers expand and tailor their Code product usage

Semgrep provides a security scanning tool that helps software teams identify vulnerabilities in code before production. It works by analyzing code with Semgrep OSS and Pro Engine and integrates into developers’ workflows and ticketing systems for actionable insights. It reduces noise by using reachability analysis to cut false positives from open-source vulnerabilities by up to 98% and achieves fast scans—average under 5 minutes with a 10-second median CI scan. Its goal is to help engineering teams ship secure software faster by continuously finding and fixing vulnerabilities during the SDLC.

Company Size

201-500

Company Stage

Series D

Total Funding

$193M

Headquarters

San Francisco, California

Founded

2017

Simplify Jobs

Simplify's Take

What believers are saying

  • Menlo Ventures leads $100M Series D in February 2025 for AI expansion.
  • Cathy Polinsky hired as co-CTO in 2025 scales hybrid LLM engineering.
  • OpenAI program selection secures Snowflake, Figma, Dropbox customers.

What critics are saying

  • Snyk siphons SCA revenue via superior GitHub and IDE integrations.
  • GitHub Copilot free scanning commoditizes Semgrep within 18-24 months.
  • Microsoft AI investments embed security in Copilot, destroying standalone demand.

What makes Semgrep unique

  • Semgrep Multimodal combines AI reasoning and rule-based analysis for 8x more vulnerabilities.
  • Semgrep cuts false positives 98% with reachability analysis on third-party dependencies.
  • Semgrep delivers scans in under 5 minutes, median CI at 10 seconds.

Help us improve and share your feedback! Did you find this helpful?

Benefits

Health Insurance

Paid Vacation

401(k) Retirement Plan

Professional Development Budget

Flexible Work Hours

Remote Work Options

Growth & Insights and Company News

Headcount

6 month growth

0%

1 year growth

2%

2 year growth

0%
The Associated Press
Mar 19th, 2026
Semgrep launches Multimodal, combining AI with rule-based analysis to find 8x more vulnerabilities

Semgrep has launched Semgrep Multimodal, a code security system combining AI reasoning with rule-based analysis for vulnerability detection, triage and remediation. The system finds up to eight times more true positives whilst cutting noise by 50% compared to foundation models alone, and has discovered dozens of zero-day vulnerabilities at customer sites. Built on Semgrep Workflows, the framework enables security teams to automate processes using deterministic tools and AI. Traditional rule-based scanners excel at catching known vulnerabilities but struggle with business logic flaws, whilst LLMs alone produce high false positive rates. Semgrep Multimodal addresses both dimensions by pairing precise programme analysis with LLM reasoning. Semgrep Multimodal is available today, with custom workflows accessible via private beta. Companies including Snowflake, Figma and Dropbox use Semgrep's platform.

Silicon Valley Journals
Feb 5th, 2025
Semgrep Raises $100M Series D Funding Round

Semgrep, a leading application security platform, has secured $100 million in Series D funding, led by Menlo Ventures with participation from existing

Semgrep
Apr 19th, 2023
Semgrep, a code & supply chain security search engine, raises Series C

Announcing our $53M Series C led by Lightspeed Venture Partners

r2c
May 11th, 2022
R2c launched DeepSemgrep for Java and Ruby on May 11th 22'.

Recognizing the value of deeper vulnerability detection, today R2c is announcing DeepSemgrep for Java and Ruby.

r2c
Oct 21st, 2021
R2c is developing Semgrep

When R2c began developing Semgrep that was its main focus, and R2c knew that lightweight static analysis, based on syntax-aware matching, would excel at enforcing secure defaults.

INACTIVE