Engineering insights from OmniDataTec

More articles

1. 

   Data Science9 min read

   AI Agents Plus Gurobi: Where the LLM Stops and the Solver Starts

   Where to draw the boundary between a language model and a MILP solver in a production decision system, and what breaks when the boundary is fuzzy.

2. 

   Software Engineering5 min read

   API vs. GUI Automation for AI Agents

   Why GUI automation is a fallback rather than a strategy: no contract, no idempotency, no error taxonomy — and the wrapper layer most teams skip.

3. 

   Data Science5 min read

   Account-Takeover Detection With Session and Device Graphs

   Account takeover is a change-of-actor problem, not an anomaly problem. How session and device graphs give you features that per-account modelling cannot.

4. 

   Artificial Intelligence6 min read

   Agent Checkpoints and Resumability for Multi-Hour Tasks

   What to write at a checkpoint, why replaying an agent is not the same as restoring it, and how to resume without re-issuing side effects or bad state.

5. 

   Artificial Intelligence5 min read

   Agent Identity and Delegated Authorisation in Enterprise AI

   Agents inherit service accounts and audit trails stop naming anyone. What agent identity has to separate, and what a delegated grant must carry.

6. 

   Architecture7 min read

   Agent Memory Creates a Data Governance Problem

   Agent memory is a personal-data store that nobody registered. What breaks — classification, provenance, purpose limitation, deletion — and how to scope it.

7. 

   Artificial Intelligence6 min read

   Agentic AI vs. Traditional Workflow Automation: A Decision Framework

   Two variables decide it: how much the input space branches, and what a wrong action costs. A framework, the hybrid that usually wins, and where each fails.

8. 

   Data Science7 min read

   Agentic Decision Intelligence for Budget Allocation

   What an agent adds to budget allocation that a scheduled solver job does not, where the autonomy dial should sit, and how the pattern fails in production.

9. 

   Artificial Intelligence9 min read

   Agentic RAG: Letting the Agent Decide When and What to Retrieve

   Five retrieval decisions you can hand to an agent, what each costs when it goes wrong, and the bounds that make adaptive retrieval safe in production.

10. 

    Software Engineering5 min read

    Agentic Testing for Financial Applications

    Agents are good at reaching states your test suite never imagined and bad at knowing whether the balance is right. Design the oracle first.

11. 

    Data Science7 min read

    Alert-Triage Agents: Cutting AML Backlogs Without Cutting Coverage

    How to use agents to clear transaction-monitoring backlogs without quietly narrowing detection coverage, and where the human decision must stay.

12. 

    Data Science7 min read

    Architecting an AI Agent for AML Alert Investigation

    A reference decomposition for an alert-investigation agent: what it gathers, what it may decide, and the failure modes that only show up after deployment.

13. 

    Software Engineering5 min read

    Architecture Patterns for AI-Generated Software

    When code becomes cheap to write, architecture should optimise for cheap verification and cheap replacement. Which patterns gain value, and which get worse.

14. 

    Machine Learning8 min read

    Are Tabular Foundation Models Finally Challenging XGBoost?

    TabPFN-2.5 reports beating default XGBoost on every small table. Read the word default carefully, then decide what to move.

15. 

    Machine Learning5 min read

    AutoML in the Foundation-Model Era: What PyCaret-Style Tools Are Still For

    Tabular foundation models changed what a baseline costs. The case for keeping a low-code AutoML layer, and the three jobs it still does better than anything.

16. 

    Machine Learning7 min read

    Automated Feature Engineering With Coding Agents: What to Delegate

    Coding agents write feature code faster than any team can review it. A gate-based delegation model for deciding what they may generate and what they may not.

17. 

    Software Engineering5 min read

    Automation at the Edges: Lessons From Running 60+ Production Bots

    Sixty-plus moderation and anomaly-detection bots taught the governance lessons the agent industry is rediscovering: idempotency, undo, and permission.

18. 

    Data Engineering7 min read

    Autonomous Data-Quality Agents: Useful Automation or Alert Noise?

    Data-quality monitoring is easy to deploy and easy to make useless. Where agents help, where the alert arithmetic defeats them, and how to budget alerts.

19. 

    Machine Learning6 min read

    Benchmark Leakage in Time-Series Foundation Model Evaluations

    Pre-training overlap inflates reported forecasting accuracy by 47-184%. How the contamination happens, and how to evaluate so it cannot.