Open to opportunities

Karhira Salzgeber

@karhirasalzgeber

Message

AI evaluation and quality specialist improving LLM accuracy at production scale.

Italy

Message

What I'm looking for

I’m looking for a remote, autonomous role where I can run fast LLM evaluation and quality audits, design benchmarks, and improve training pipelines by finding systemic errors—while collaborating with teams that value rigor, consistency, and real-world data cleanliness.

I’m an AI Operations & Quality specialist focused on evaluating LLM outputs at production scale, catching systemic failures that others miss. I’ve worked across multiple evaluation divisions, including video RLHF, LLM RLHF annotation & eval, world knowledge benchmarks, intent understanding, emotion comprehension, and preference ranking.

I move fast without cutting corners and pick up new workflows with minimal hand-holding. After consistently hitting the top of the internal quality grading system, I took on reviewer responsibilities at Anuttacon/Verity Lab, auditing annotator submissions for accuracy, consistency, and compliance, and contributing to workflow corrections when pattern failures and annotation errors showed up in live training pipelines.

Experience

Work history, roles, and key accomplishments

Current

AI Operations & Quality Specialist

Current

Anuttacon / Verity Lab

Jun 2025 - Present (1 year 1 month)

Work across multiple LLM and video evaluation divisions, including reviewing outputs against RLHF rubrics, designing benchmark questions, and running preference ranking and reward modelling. Also perform workflow QA by auditing submissions, calibrating annotation, and identifying edge-case/systemic issues.

Quality Assurance Benchmark Design Question Writing Error & Edge Case Identification Annotation Calibration & Feedback World Knowledge Benchmarks Compliance Auditing Slack Microsoft Teams Google Workspace Microsoft Office Suite

AI Trainer & Quality Specialist

Invisible Technologies

Aug 2024 - Apr 2025 (8 months)

Train and evaluate AI models using structured annotation and prompt review, then assess outputs for factual accuracy, tone, and instruction alignment. Build and run benchmarks and scoring (including humanised preference/reward modelling) while contributing to internal quality efforts and workflow fixes.

Prompt Review LLM Evaluation Evaluating Factual Accuracy World Knowledge Benchmarks Reality TV Benchmarks Documentary Benchmarks Travel Benchmarks Accommodation Benchmarks Humanised Preference Scoring Annotation Error Identification

Freelance Civil Engineer

Grey Studios Ltd

Jul 2019 - Apr 2023 (3 years 9 months)

Managed project timelines, budgets, and contractors for residential renovation work and produced/edited technical documentation and client reports to a high standard of accuracy. Ran data analysis and client-facing reporting to keep projects on track with strategic goals.

Construction Project Management Budgeting Contractor Management Residential Renovation Delivery Technical Documentation Client Reporting Data Analysis Strategic Reporting

General Foreman / Junior Engineer

S&B Construction

May 2016 - Jun 2019 (3 years 1 month)

Oversaw site logistics and coordinated teams across multiple active builds while working with engineers and trades to deliver work on schedule. Conducted safety analyses and also performed annotation alignment and judgment calibration for LLM evaluation as part of training/self-study.