Open to opportunities

Pooyan Alavi

@pooyanalavi

Message

AI Software Engineer building grounded LLM systems and low-latency recommendation & chatbot APIs.

Armenia

Message

What I'm looking for

I’m looking to build AI products where grounded LLMs, hybrid retrieval, and low-latency serving matter—measuring p95/cost, iterating with real evaluation, and owning systems that scale.

I’m an AI Software Engineer and co-founder/CTO working on production AI systems—from hybrid recommendation to grounded LLM experiences. At Ontime App, I built a hybrid recommendation engine combining PostgreSQL and Neo4j, synchronized via Apache Kafka so both stores stayed eventually consistent without blocking the API.

I also developed recommendation and chatbot APIs with FastAPI and Redis caching, cutting p95 latency by ~40%, and fine-tuned an open-source LLM on a local corpus with a vector RAG pipeline to reduce hallucinations and reliance on costly external models. Earlier roles strengthened my end-to-end craft: I built an MLflow-tracked data/ML pipeline for generating tailored Instagram content for 20 franchises, used the ELK stack for KPI monitoring, and deployed semantic search models on AWS (SageMaker, S3, EC2) with Redis for low-latency inference.

Experience

Work history, roles, and key accomplishments

Current

Co-Founder & CTO

Current

Trusty

Mar 2026 - Present (4 months)

Co-founded Trusty and served as CTO, leading the company’s technology direction.

Current

AI Software Engineer

Current

Ontime App

Feb 2025 - Present (1 year 5 months)

- Built a hybrid recommendation engine combining PostgreSQL (relational user/item data) with Neo4j (graph database), kept in sync through a Kafka message queue. Chose a graph database to traverse multi-hop user–community relationships that relational joins handled inefficiently, and used Kafka to decouple ingestion from serving so both stores stayed eventually consistent without blocking the API.

Knowledge Engineering LLM Neo4j GNN Knowledge Graphs Kafka PostgreSQL fastAPI Redis Embeddings Recommendation Systems

Senior Data Engineer

CoLife Labs

Mar 2024 - Feb 2025 (11 months)

- Built an MLflow-tracked pipeline that extracted market trends and generated tailored Instagram content for 20 franchises of the lab.
- Applied classic ML (XGBoost) and data processing (cleaning, ETL, embedding-based semantic search) to surface trends from noisy, multi-source data.
- Used the ELK stack (Elasticsearch, Kibana) to monitor KPIs and report to stakeholders, and optimized MongoDB queri

Software Development MLFlow XGBoost ETL Data Cleaning Elasticsearch Kibana MongoDB KPIs

Senior Data Scientist

RoboEpics

Jun 2022 - Aug 2023 (1 year 2 months)

Pretrained a BERT model on a local corpus to power domain-specific semantic search and embeddings.
Researched state-of-the-art ML and designed AI competitions, collaborating with the CEO to shape new AI products.
Trained and deployed models on AWS (SageMaker, S3, EC2) with Redis caching for low-latency inference.