Beacon Venture Capital
Beacon Venture Capital
Beacon Venture Capital is the wholly owned corporate venture capital arm of Kasikornbank PLC, a leading commercial bank in Thailand. Established in 2016, the fund invests strategically in technology startups from early through growth stages, with a primary focus on fintech. Its investment interests also include consumer lifestyle businesses and deep technology areas such as artificial intelligence and enterprise IT. Beacon Venture Capital launched with $30 million in capital and grew to $135 million by 2018.

Senior Software Engineer, Artificial Intelligence and LLM

Build and operate production LLM capabilities for Beacon AI’s aviation safety platform. Lead retrieval, tool calling, evaluation, and reliable service delivery.

Description

  • Build retrieval-augmented generation and tool-calling workflows
  • Produce validated JSON and schema-constrained outputs with retries and fallback handling
  • Connect function calling to internal tools, search, routing, and data services
  • Develop Python or TypeScript APIs and workers with defined contracts, streaming, and backoff
  • Use caching, request shaping, prompt templates, and context packing to manage latency and costs
  • Integrate foundation model services from AWS Bedrock, OpenAI, Anthropic, and self-hosted endpoints
  • Help develop chunking, embedding, and indexing strategies for documents, time series, and multimedia
  • Select and optimize vector databases including OpenSearch, pgvector, and Pinecone
  • Keep knowledge bases synchronized with S3, Aurora, DynamoDB, and external data sources
  • Develop offline evaluations and golden datasets for prompts, retrievers, and tools
  • Define production metrics covering task success, hallucination rate, retrieval precision and recall, p95 latency, and cost per request
  • Conduct A/B tests and manage guarded prompt and version releases with canary deployments
  • Implement content and policy safeguards, PII detection and redaction, access controls, and audit trails
  • Create human-review workflows for sensitive actions
  • Instrument model calls with tracing, logging, and dashboards for token usage, errors, and saturation
  • Investigate issues across retrieval systems, prompts, tools, and model providers
  • Work with product, infrastructure, ML, and security teams to deliver dependable aviation systems
  • Create user-facing LLM capabilities for Beacon AI’s platform to make flying safer, more efficient, and more capable

Requirements

  • 5–8 years of experience, including experience delivering production ML or LLM systems
  • Demonstrated experience launching LLM features for users and improving them through data
  • Strong production software development, testing, and documentation skills
  • Working knowledge of embeddings, chunking, vector search tradeoffs, and function calling
  • Experience creating evaluations, defining success measures, and iterating from evidence
  • Ability to monitor p95 latency, meet SLAs, and lower costs while maintaining quality
  • Ability to take ownership of features from design through production with limited supervision
  • Experience with Bedrock, OpenSearch Serverless, pgvector, Pinecone, or Weaviate is preferred
  • Experience with prompt versioning, guardrails, and model-provider routing is preferred
  • Experience with multimodal data such as time series or video is preferred
  • Familiarity with GPU inference, Triton, or TensorRT-LLM is preferred
  • Exposure to aviation or another safety-critical industry is preferred
  • Working knowledge of CI/CD, infrastructure as code, and secure secrets management is preferred
  • Must qualify as a U.S. Person
  • Must be legally authorized to work in the United States
  • Visa sponsorship and visa transfers are not available
  • All work must be performed within the United States
  • Must live in or be willing to relocate to the San Francisco Bay Area
  • Must work onsite in San Carlos, California, at least three days per week
  • Must be available for full-time work

Benefits

  • Equity is included in the offer
  • The company covers 100% of employee medical premiums
  • The company covers 25% of dependent medical premiums
  • Three weeks of paid time off
  • At least 13 paid company holidays
  • 401(k) plan available
  • Flexible remote work on the remaining days of the hybrid schedule

Related Jobs

Mindrift

Senior Python Data Scraping Engineer (Freelance, Remote in Sweden)

Mindrift

Mindrift is seeking a senior Python engineer to build dependable scraping workflows for AI projects. The role focuses on extracting, validating, and scaling structured data from complex, dynamic websites.

Open
Mindrift

Senior Python Data Scraping Engineer (Freelance, Remote in Saudi Arabia)

Mindrift

Senior Python engineer developing scalable scraping workflows for Mindrift’s Tendem AI project. The role focuses on extracting, validating, and structuring complex web data with Python and automation tools.

Open
Mindrift

Senior Python Data Scraping Engineer — Freelance Remote in Romania

Mindrift

Mindrift is seeking a senior Python engineer to build dependable scraping and data-extraction workflows for AI projects. The role focuses on converting dynamic websites into accurate, validated, structured datasets.

Open
Mindrift

Senior Python Data Scraping Engineer — Remote in the Philippines

Mindrift

Mindrift is seeking a senior Python engineer to build dependable scraping workflows for AI projects. The role focuses on extracting, validating, and scaling structured data from complex, dynamic websites.

Open
Cisco

Senior Business Development Manager, Splunk National Security — Cisco

Cisco

Lead strategic business development and capture efforts for Cisco and Splunk cybersecurity, analytics, and AI solutions supporting U.S. National Security programs. Shape federal opportunities across Intelligence Community and classified government accounts.

Open