Mohammad Akbari Monfared

Hamburg, DE

• Led the full product lifecycle for two live AI products at DCV, owning design, implementation, delivery, and alignment with non-technical founders and business teams.
• Built and shipped AIME (aimetalent.com), an agentic RAG product that cut a multi-day manual recruitment process to minutes by automating global sourcing and matching, now used daily by recruiters.
• Engineered an agentic CV Matcher using RAG and deep semantic search, delivering high-precision candidate ranking and discovery across large databases.
• Built an automated ETL pipeline that ingests CVs from a data lake, routes them by detected location, and uses LLMs to extract and structure profiles before loading them into vector and document databases (FAISS, MongoDB), triggered end-to-end without manual steps.
• Designed and shipped a recruitment CRM with a team of two developers, covering the full lifecycle from client contact to commission payout by automating invoicing, margin calculation, and contract generation.
• Owned cloud and deployment for both products on AWS, building CI/CD pipelines with Docker for automated testing and deployment to EC2 across staging and production, with data storage on S3, MongoDB, and RDS.

Mönchengladbach, DE

• Built and shipped an AI product for cold-rolling mills, taking it from idea to a system used in factories today, increasing production efficiency by 2%.
• Designed an ETL pipeline processing high-frequency sensor data (10GB per 10-minute window), extracting compact, high-value summaries that resolved storage bottlenecks while preserving critical information.
• Developed a custom statistical forecasting and anomaly-detection model as the core of the product, outperforming standard benchmarks by 30%.
• Deployed end-to-end with integration into factory systems via gRPC and SQL, delivering real-time dashboards in Grafana that served predictions and monitored model performance and data drift.
• Partnered with domain experts and senior management to translate operational challenges into the product roadmap and retraining logic that improved plant-level productivity.

Bonn, DE

• Developed the first agentic framework for Aspect-Based Sentiment Analysis (ABSA), using an LLM "Generator-Evaluator" (LLM-as-judge) workflow to produce label-consistent synthetic data; published at EACL 2026.
• Closed the performance gap for lightweight models with self-reflection logic, yielding a 6% F1-score improvement over standard augmentation.
• Designed a systematic evaluation framework across four SemEval benchmarks, proving agentic augmentation can match real-world data performance in low-resource scenarios.
• Engineered the end-to-end Python pipeline studying task complexity and data ratios on downstream ML performance for data-scarce training strategies.

Remote

• Developed and maintained Python scripts automating internal data reporting workflows, saving the operations team several hours per week with repeatable pipelines.
• Built and consumed RESTful API endpoints, working across request/response cycles, authentication flows, and backend service integration.
• Queried and maintained SQL databases as part of feature delivery, developing strong proficiency in structured data systems and schema design.
• Contributed to a shared codebase using Git, following code review processes and resolving merge conflicts independently.