
Researcher, LLM Evaluation — AMALIA Programme
Lisbon
Wigi highlights
1+ year, currently working as Junior Researcher
Built AMALIA's reasoning-judging stage from scratch
Ran reproducible SLURM benchmarking across 20+ configurations
Custom vLLM and SGLang serving paths kept long reasoning traces inside budgets of 16k to 64k tokens, with GPU memory utilisation tuned up to give the KV cache enough headroom for traces that long; part of the judging ran through a commercial API where structuring the prompts around a shared prefix put ~98% of input tokens in cache and cut the bill accordingly.
In an earlier role, they also built forecasting models for unit production time, machine downtime and defect counts from raw textile process data. Separate work included benchmarking and optimising four distributed NoSQL databases (MongoDB, Cassandra, Aerospike, Redis) in C#/.NET.
Available now
Ready to start immediately. No notice period. Actively looking.
Skills
Work experience
Unlocks when they connect with you.
- projects· 4
- certifications· 2
- education· 3
- Files· 1
Similar available Generative AI & LLM Engineers
View all- AI Platform Engineer | Vector Search & Enterprise Search · Lahore, Pakistan · 4 Yrs Exp · Remote, Hybrid, On-site · $2,500/mo
- AI Content Generation Specialist · Paris, France · 8 Yrs Exp · Remote · $5,800/mo
- Prompt Engineer - Ai Trainer (SQL, Tableau) · Glasgow, UK · 2 Yrs Exp · Remote · $7,200/mo
- Synthetic Data Engineer, Génération De Cas De Test Automatisés · Paris, France · 1 Yr Exp · Hybrid · $5,600/mo
- Generative AI Engineer, AI-Driven Digital Experiences · Deva, Romania · 8 Yrs Exp · Hybrid · $3,300/mo