(612) 555-0171 ◇ Minneapolis, MN
ingmar.solvik@example.com ◇ linkedin.com/in/ingmar-solvik ◇ ingmarsolvik.com
Search engineer with 7 years building search and ranking systems, currently working on product search for a home improvement retailer whose site handles 3 million searches a day. Raised search conversion 11% with a learning-to-rank model, cut zero-result searches from 9% to 2.5%, and kept p95 search latency under 120 ms through peak season. Balances relevance, speed, and business rules so shoppers find what they came for.
Focus on Information Retrieval. Coursework in Search Engines, Machine Learning, Natural Language Processing, and Distributed Systems.
Senior project on spelling correction for product search queries.
- Work on product search for a retail site handling 3 million searches a day across 1.2 million products.
- Raised search conversion 11% with a learning-to-rank model trained on 18 months of click data.
- Cut zero-result searches from 9% to 2.5% with query rewriting and 4,000 curated synonyms.
- Built job search ranking for a job board with 600,000 live listings and 2 million monthly users.
- Raised job application starts from search 14% by adding location and skill matching signals.
- Moved search from Solr 6 to Solr 8 with 0 downtime across 3 data centres.
- Built a test collection of 5,000 judged query-document pairs for e-commerce search research.
- Co-authored 1 workshop paper on query reformulation for product search.
- Taught 2 lab sections of an information retrieval course for about 50 students.
Learning-to-Rank Rollout. Built a ranking model with XGBoost on 40 features from 18 months of clicks and purchases, served as a Solr re-ranker, tested in 3 A/B rounds, which raised search conversion 11% and added about $40M in yearly online sales.
Zero-Result Reduction. Analysed 2 million zero-result queries, built a query rewriting service with spelling correction and 4,000 curated synonyms, and added a vector search fallback, which cut zero-result searches from 9% to 2.5%.
Peak Season Performance. Load tested search at 3 times normal traffic, added result caching and tuned index sharding, which kept p95 latency under 120 ms through peak season with 0 search outages.
- Speaker on e-commerce relevance at 2 search technology conferences since 2023.
- Contributor to an open source learning-to-rank plugin, with 8 merged pull requests.
- Organise a Minneapolis search and relevance meetup with about 300 members.
- Technical lead for ranking on a search team of 6 engineers and 1 data scientist.
- Run a weekly relevance review with merchandising and product teams on 20 key queries.
- Mentor 2 junior engineers on search relevance and A/B test analysis.

.webp)



.webp)


%20Which%20Should%20You%20Use.webp)