Mihnea is available for hire
Hire MihneaMihnea Dinu
Verified Expert in Engineering
Prompt Engineering Developer
Bucharest, Romania
Toptal member since September 29, 2020
Bio
Mihnea is an AI engineer specializing in agentic AI for regulated environments, with eight years of experience in financial services at ING, Inspari, and Commonwealth Financial Network. He architects stateful multi-agent systems using LangGraph and LangChain that provide consistent, explainable answers. Mihnea recently led Commonwealth through its first LangGraph production deployment in Databricks, reducing reporting from three days to three minutes while meeting governance requirements.
Portfolio
Commonwealth Financial Network - Main
Microsoft Power BI, DAX, Business Intelligence (BI), Databricks, Snowflake...
Inspari A/S
SQL, Microsoft Power BI
ING Tech
SQL Server Integration Services (SSIS), SQL Server Analysis Services (SSAS)...
Experience
- Business Intelligence (BI) - 8 years
- SQL - 5 years
- Prompt Engineering - 1 year
- Generative Artificial Intelligence (GenAI) - 1 year
- LangGraph - 1 year
- LangChain - 1 year
- Large Language Models (LLMs) - 1 year
- Agentic AI - 1 year
Preferred Environment
Windows
The most amazing...
...thing I’ve built is a production conversational analytics agent that passed governance reviews and reduced reporting time from three days to three minutes.
Work Experience
Senior Business Intelligence Consultant | AI Engineer
2021 - PRESENT
Commonwealth Financial Network - Main
- Led production deployment of a conversational AI agent using LangGraph and LangChain in Databricks. Built a stateful multi-agent system maintaining conversation context across sessions.
- Architected a decision-safety framework with orchestrator routing questions into scenarios (data query, pleasantries, unavailable, and ambiguous). Added a clarification step before query generation and a transparency layer surfacing assumptions.
- Created an evaluation framework scoring answers across five dimensions: Clarification, No Data Acknowledgment, Accuracy, Interpretation, and Transparency. Used LLM judges to evaluate golden datasets and production answers.
- Implemented observability tracking agent answers and intermediate steps using structured outputs. Built monitoring for SQL query result sizes and conversation history token limits for debugging.
- Reduced reporting time from three days to three minutes for executives, eliminating 3,600 manual hours annually. The agent passed the governance review by enforcing organizational language aligned with the revenue statement.
- Increased user engagement from 15% to 40% in six months by revamping the digital platform. Built Power BI dashboards integrating CRM and Google Analytics data for user segmentation and drop-off analysis.
- Helped the company meet 90% of SLA targets without extra hires. Built end-to-end analytics for 1+ million support cases, uncovering process bottlenecks and enabling smarter case distribution.
Technologies: Microsoft Power BI, DAX, Business Intelligence (BI), Databricks, Snowflake, Large Language Models (LLMs), LangChain, LangGraph, Prompt Engineering, Generative Artificial Intelligence (GenAI), Python, SQL, Agentic AI
Business Intelligence Consultant
2019 - 2021
Inspari A/S
- Worked as a business intelligence consultant in different lifecycles of the projects, such as gathering requirements, implementing data models and reporting solutions, quality assurance, and maintaining the quality of deliverables.
- Projected the amount of money lost for a retail company for all goods during transit or delay in arrival. Gathered information from the customer about the desired solution, planned the deliverables and implementation, and did quality assurance.
- Migrated multidimensional cubes to tabular models for the insurance industry with reporting requirements for budgets. I migrated OLAP data models to tabular data models.
- Designed the database architecture and implemented a data warehouse for reporting national sports activities. Created a file management solution to centralize all information in the database model. Added business logic to meet reporting requirements.
- Created reporting solutions based on Power BI premium licenses. I developed the source queries for extracting data from the SQL data warehouse and Azure Databricks, the tabular model, and the report. Technology stack: Power BI.
- Involved in the front-end and UX development in Power BI: gathered reporting requirements from the product owner and adjusted the report interface to make it user-friendly.
Technologies: SQL, Microsoft Power BI
Business Intelligence Developer
2017 - 2019
ING Tech
- Provided DevOps maintenance for regulatory reporting, tracking file management, and loading.
- Implemented a file-based reporting workflow based on finance data sources: Bloomberg, Sophis, and Summit.
- Integrated all data sources in tabular and multidimensional cubes.
Technologies: SQL Server Integration Services (SSIS), SQL Server Analysis Services (SSAS), SSAS Tabular
Business Intelligence Developer
2015 - 2017
Bearing Point
- Developed an HR solution to track the performance of consultants.
- Created sales pipeline reporting to show projects with the most promising leads.
- Designed balance sheet and profit/loss reporting for every quarter for top management.
Technologies: SQL
Experience
Enterprise Analytics Agent: Production LangGraph Deployment
https://github.com/mdinu-hash/analytics_agent_databricksI developed a production-grade conversational analytics agent for the financial services industry.
ARCHITECTURE
• Multi-agent orchestration using LangGraph and conditional routing
• Stateful workflows tracking conversation history and intermediate execution steps
• Tool calling with orchestrator, labeling questions into scenarios, and routing to appropriate tools
• ChatDatabricks integration with structured outputs for reliable parsing
DECISION SAFETY FEATURES
• Disambiguation asking clarifying questions ("Which revenue—gross, net, recurring?")
• Revealing assumptions (e.g., "Revenue: $5 million–excludes inactive accounts")
• Schema-grounded analysis preventing hallucinations
• Clarification step before query generation handling analytical intent ambiguity
PRODUCTION QUALITY
• Evaluation framework scoring across five dimensions: Clarification, No Data Acknowledgment, Accuracy, Interpretation, and Transparency
• Observability tracking all answers and intermediate steps for debugging
RESULTS
• Reduced reporting cycles from three days to three minutes
• Eliminated up to 3,600 manual hours annually
• Consistent answers across contexts
ARCHITECTURE
• Multi-agent orchestration using LangGraph and conditional routing
• Stateful workflows tracking conversation history and intermediate execution steps
• Tool calling with orchestrator, labeling questions into scenarios, and routing to appropriate tools
• ChatDatabricks integration with structured outputs for reliable parsing
DECISION SAFETY FEATURES
• Disambiguation asking clarifying questions ("Which revenue—gross, net, recurring?")
• Revealing assumptions (e.g., "Revenue: $5 million–excludes inactive accounts")
• Schema-grounded analysis preventing hallucinations
• Clarification step before query generation handling analytical intent ambiguity
PRODUCTION QUALITY
• Evaluation framework scoring across five dimensions: Clarification, No Data Acknowledgment, Accuracy, Interpretation, and Transparency
• Observability tracking all answers and intermediate steps for debugging
RESULTS
• Reduced reporting cycles from three days to three minutes
• Eliminated up to 3,600 manual hours annually
• Consistent answers across contexts
Achieving 90% SLA Compliance Without New Hires Through Workflow Analytics
I designed a Workflow Analytics end-to-end analysis (using data from Snowflake and Azure Data Factory) to analyze over one million support cases. The report helped operations leaders identify hidden process bottlenecks and workload imbalances across teams. Insights led to targeted cross-training and smarter case distribution, allowing the company to meet 90% of SLA expectations without hiring additional staff. This solution enabled sustainable growth while protecting service excellence.
Boosting User Retention from 15% to 40% with Customer Analytics
Developed a centralized Customer Insights BI report using Power BI, Snowflake, and Azure Data Factory. By unifying CRM and Google Analytics data, we enabled leadership to identify usage patterns, segment user engagement, and uncover key drop-off points. The insights reshaped the product strategy, shifting focus to client acquisition and retention, resulting in a 3x increase in regular users within six months.
Education
2015 - 2015
Exchange Program Participant in Environmental Sciences
University of New South Wales - Sydney, Australia
2012 - 2015
Master's Degree in Environmental Sciences
Technical University of Hamburg - Hamburg, Germany
Certifications
JANUARY 2021 - JANUARY 2024
Power BI Data Analyst Associate
Microsoft
Skills
Tools
Microsoft Power BI
Paradigms
Database Design, Business Intelligence (BI)
Languages
SQL, Python, Snowflake
Frameworks
LangGraph
Platforms
Databricks
Other
Data Modeling, DAX, LangChain, Generative Artificial Intelligence (GenAI), Prompt Engineering, Large Language Models (LLMs), AI Product Management, Azure Data Factory (ADF), Agentic AI
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring