Hi, I’m Edgaras Jeparchinas, an AI Engineer focused on autonomous agent systems across finance, surveillance, and enterprise intelligence. I design and deploy full‑stack ML/LLM ecosystems, including scalable data pipelines, vector retrieval, cloud infrastructure, agent orchestration, and production‑grade deployment. I enjoy building modular RAG platforms with automated cloud provisioning, FAISS/pgvector indexing, recursive chunking, and governance layers. My work spans finance underwriting agents, simulation pipelines, surveillance and asset validation systems, and large‑scale NLP/LLM extraction workflows. I also run adversarial testing (Garak, TextAttack, IBM ART) to enhance robustness, fairness, and jailbreaking resistance, while ensuring auditable, compliant AI systems in AWS and Azure.

Edgaras Jeparchinas

Hi, I’m Edgaras Jeparchinas, an AI Engineer focused on autonomous agent systems across finance, surveillance, and enterprise intelligence. I design and deploy full‑stack ML/LLM ecosystems, including scalable data pipelines, vector retrieval, cloud infrastructure, agent orchestration, and production‑grade deployment. I enjoy building modular RAG platforms with automated cloud provisioning, FAISS/pgvector indexing, recursive chunking, and governance layers. My work spans finance underwriting agents, simulation pipelines, surveillance and asset validation systems, and large‑scale NLP/LLM extraction workflows. I also run adversarial testing (Garak, TextAttack, IBM ART) to enhance robustness, fairness, and jailbreaking resistance, while ensuring auditable, compliant AI systems in AWS and Azure.

Available to hire

Hi, I’m Edgaras Jeparchinas, an AI Engineer focused on autonomous agent systems across finance, surveillance, and enterprise intelligence. I design and deploy full‑stack ML/LLM ecosystems, including scalable data pipelines, vector retrieval, cloud infrastructure, agent orchestration, and production‑grade deployment.

I enjoy building modular RAG platforms with automated cloud provisioning, FAISS/pgvector indexing, recursive chunking, and governance layers. My work spans finance underwriting agents, simulation pipelines, surveillance and asset validation systems, and large‑scale NLP/LLM extraction workflows. I also run adversarial testing (Garak, TextAttack, IBM ART) to enhance robustness, fairness, and jailbreaking resistance, while ensuring auditable, compliant AI systems in AWS and Azure.

See more

Experience Level

Expert
Expert
Expert
Expert
Expert
Expert
Expert
Expert
Expert
Intermediate
Intermediate
See more

Language

English
Fluent

Work Experience

Gen AI Data Scientist (Contract) at Infinigate Group
December 1, 2025 - December 1, 2025
Designed and deployed a modular RAG platform with automated Azure infrastructure, unified knowledge bases, agent interfaces, and evaluation systems for rapid, scalable deployment of high-accuracy AI agents. Built a high-performance ingestion engine with recursive chunking and semantic retrieval (FAISS/pgvector). Implemented vendor-agnostic document normalization and integrated Azure OpenAI embeddings with RAG profiles and isolation-based reproducibility. Led the Agent Judge evaluation pipeline to audit outputs for hallucinations, retrieval consistency, and citation accuracy, plus a CLI/Gradio interface for multi-agent orchestration and monitoring.
AI Engineer at Climate X
July 1, 2025 - July 1, 2025
Optimized legacy systems and designed next-generation surveillance solutions. Built a scalable computer vision pipeline (YOLO/OpenCV) for asset classification with geospatial overlays and cross-checks. Developed an LLM-guided web-scraping agent to extract and structure company information, and created a geospatial asset validation pipeline with rule-based audits. Led adversarial red-teaming exercises to ensure safety, bias control, and regulatory alignment.
Remote Data Scientist at Propel Finance
January 31, 2024 - January 31, 2024
Led AI-driven underwriting and market simulation platforms, including sentiment analytics, credit decision models, and 360-degree customer profiling. Implemented a Q-network reinforcement layer for dynamic PD modeling, end-to-end Azure Databricks/Data Lake pipelines, and SHAP explainability for transparent credit decisions. Built Baseline pre-screening using Companies House data to reduce manual underwriting costs while maintaining regulatory auditability.
Hybrid Senior Python Developer at LCH
June 1, 2022 - June 1, 2022
Led development of a unified operational platform to streamline system communication, increase resilience, and reduce costs. Rebuilt cost-valuation architecture, developed anomaly detection (cGANs, Q-networks), and contributed to pricing, stress testing, and regulatory reporting. Collaborated across regions to automate cost controls and data reconciliation.
Hybrid Quantitative Analyst at UBL UK
August 31, 2019 - August 31, 2019
Streamlined risk management and regulatory modeling. Built IRRBB stress tests, IFRS9 provisioning, AFS/HTM monitoring, Net FX modeling, and consumer/CRE lending risk models with integrated stress testing and governance. Contributed to model validation, regulatory reporting, and cross-asset risk analytics.

Education

MSc Artificial Intelligence at University of Leeds
January 1, 2021 - January 1, 2023
BSc Accounting and Finance at Coventry University London
January 1, 2015 - January 1, 2018

Qualifications

Add your qualifications or awards here.

Industry Experience

Financial Services, Software & Internet, Media & Entertainment, Professional Services, Other