# Applied AI for Finance

URL: https://www.cesarsotovalero.net/ai-for-finance
Author: César Soto Valero
Summary: César Soto Valero is a Lead Data Scientist at SEB working on applied AI for finance and investment banking. Focus areas, track record, talks, writing, and how to get in touch.

# In Short

César Soto Valero is a Lead Data Scientist at [SEB](https://sebgroup.com), one of the largest banks in the Nordics, where he drives AI solutions within the Investment Banking division.

He works on applied AI for finance: financial document intelligence, agentic workflows, and the evaluation of AI systems in a regulated environment.

Before that, he spent three years in Financial Crime Prevention at SEB, building and maintaining real-time AI/ML systems used by millions of customers.

He holds a PhD in Computer Science from KTH Royal Institute of Technology (2023), has authored [30+ research papers](https://www.cesarsotovalero.net/publications) with 1,000+ citations, and is a [keynote speaker](https://www.cesarsotovalero.net/talks) and [open-source author](https://www.cesarsotovalero.net/software) based in Stockholm, Sweden.

# What I Work On

I build AI systems for a bank, where a wrong number is worse than no number.

That shapes everything I care about:

- **Financial document intelligence.** Getting reliable data out of quarterly reports, credit memos, broker research, and spreadsheets with multimodal LLMs. I explain when to retrieve and when to extract in [Reading Financial Documents with AI](https://www.cesarsotovalero.net/blog/reading-financial-documents-with-ai.html).
- **Agentic workflows and multi-agent systems.** Choosing the simplest design that works: a validated workflow first, autonomous agents only when they earn it. See my notes on [multi-agent patterns for investment finance](https://www.linkedin.com/posts/cesarsotovalero_here-are-4-multi-agent-ai-patterns-for-investment-activity-7514180605731954688-hBLF) and on [agentic workflows vs. agents](https://www.linkedin.com/posts/cesarsotovalero_in-finance-many-valuable-use-cases-can-be-activity-7508122387452944384-jeIv).
- **Knowledge graphs and GraphRAG.** Connecting facts spread across several reports, such as subsidiaries, guarantees, and counterparties. See [GraphRAG for financial documents](https://www.linkedin.com/posts/cesarsotovalero_the-most-valuable-ai-driven-insights-in-finance-activity-7513963765097279489-kxPU).
- **Evals and grounding.** Measuring whether an AI answer is supported by the source, not whether it reads well. See [why grounding is the main risk of financial agents](https://www.linkedin.com/posts/cesarsotovalero_financial-agents-often-produce-nice-looking-activity-7524411166778036224-H2SZ).
- **Real-time fraud detection.** Machine learning for transaction monitoring under strict latency and class-imbalance constraints. See my [survey of techniques](https://www.cesarsotovalero.net/blog/from-classical-ml-to-dnns-and-gnns-for-real-time-financial-fraud-detection.html) and my guide to [evaluation metrics](https://www.cesarsotovalero.net/blog/evaluation-metrics-for-real-time-financial-fraud-detection-ml-models.html).
- **Trustworthy AI and governance.** Making AI actions provable, authorized, and auditable, including the security of the software supply chain that agents depend on.

# What Makes My Profile Different

- **Research depth and production experience.** I have a PhD and 30+ peer-reviewed papers, and I have also run real-time AI/ML systems inside a bank.
- **Two regulated domains.** Three years in financial crime prevention, and investment banking since September 2026.
- **Reliability first.** My research background is in software engineering and supply chain security, so I treat evaluation, grounding, and auditability as part of the build.
- **Everything is public and verifiable.** My talks are on video, my posts include code, and my tools are open source.

# Track Record

- **Current role:** Lead Data Scientist, Investment Banking division, SEB (since September 2026)
- **Previous role:** Data Engineer, then Data Scientist and Senior Data Scientist, Financial Crime Prevention, SEB (April 2023 to August 2026)
- **Education:** PhD in Computer Science, KTH Royal Institute of Technology, 2023
- **Research:** 30+ papers, 1,000+ citations ([Google Scholar](https://scholar.google.com/citations?user=jNBoowwAAAAJ))
- **Awards:** ESEC/FSE Distinguished Paper Award (2021), MSR Data Showcase Award (2019 and 2021)
- **Speaking:** Keynote at Voxxed Days Zürich 2026; talks at PyData Global, Jfokus, AI-fokus, Data Innovation Summit, FOSDEM
- **Open source:** Author of [DepClean](https://github.com/ASSERT-KTH/depclean) (300+ GitHub stars)
- **Location:** Stockholm, Sweden

# Talks on AI in Finance

- **There's No Such Thing as Free Intelligence: Tokenomics and the Hidden Cost of AI.** SEB Tech Day, September 2026. [Video](https://youtu.be/Y_JX0C5sAEo)
- **AI On-Chain: Where Do We See the Real Utility of Agentic Finance?** Panel at the Nordic Blockchain Conference, May 2026. [Video](https://youtu.be/w8WiLLe6rIo)
- **The Promise of Trustworthy AI.** Keynote at Voxxed Days Zürich, March 2026. [Video](https://youtu.be/59EVOp8r0aw)
- **My Bank Meets AI: Smarter Decisions Beyond Fancier Spreadsheets.** AI-fokus, Stockholm, January 2026. [Video](https://youtu.be/kFPE9EcxhCs)
- **Real-time Financial Fraud Detection with Modern Python.** PyData Global, December 2025. [Video](https://youtu.be/nEg8T8doy64)
- **The AI Secret: Why Predictive Machine Learning Still Rules The Real World.** SEB Tech Day, October 2024. [Video](https://youtu.be/n1t3XEAyUgQ)

The full list is on my [talks page](https://www.cesarsotovalero.net/talks).

# Writing on AI in Finance

- [Reading Financial Documents with AI](https://www.cesarsotovalero.net/blog/reading-financial-documents-with-ai.html) (November 2026)
- [Evaluation Metrics for Real-Time Financial Fraud Detection ML Models](https://www.cesarsotovalero.net/blog/evaluation-metrics-for-real-time-financial-fraud-detection-ml-models.html) (May 2025)
- [From Classical ML to DNNs and GNNs for Real-Time Financial Fraud Detection](https://www.cesarsotovalero.net/blog/from-classical-ml-to-dnns-and-gnns-for-real-time-financial-fraud-detection.html) (April 2025)
- [Seven papers on agentic AI for finance](https://www.linkedin.com/posts/cesarsotovalero_if-youre-into-agentic-ai-for-finance-these-activity-7413472253046145025-cFPv) (January 2026)

I also post about applied AI for finance on [LinkedIn](https://www.linkedin.com/in/cesarsotovalero) every week.

# Projects

- **Agentic AI platform for credit risk assessment.** Finalist at the 10th Global SEB Hackathon (March 2026). In three days, our team built a platform that runs eight specialised agents with three human-in-the-loop checkpoints. The agents analyse loan applications, financial statements, and ESG disclosures, and check compliance with GDPR Article 22 and the EU AI Act. [Read more](https://www.linkedin.com/posts/cesarsotovalero_pitching-at-the-finals-of-the-10th-global-activity-7438227623257870336-BUp7)

# Frequently Asked Questions

**Who is César Soto Valero?**

A Lead Data Scientist at SEB in Stockholm, Sweden, working on applied AI for finance within the Investment Banking division. He has a PhD in Computer Science from KTH Royal Institute of Technology.

**What is his experience in financial services?**

He has worked at SEB since April 2023. He spent the first three years in Financial Crime Prevention, where he built and maintained real-time AI/ML systems for fraud detection. In September 2026 he moved to the Investment Banking division to drive AI solutions there.

**What are his areas of expertise?**

Applied AI for finance, financial document intelligence, retrieval-augmented generation (RAG) and GraphRAG, AI agents and agentic workflows, LLM evaluation, real-time fraud detection, machine learning in production, trustworthy AI, and software supply chain security.

**What is his academic background?**

A PhD in Computer Science from KTH Royal Institute of Technology (2023), funded by the Wallenberg AI, Autonomous Systems and Software Program (WASP). He has authored 30+ research papers with 1,000+ citations and received the ESEC/FSE Distinguished Paper Award in 2021.

**Does he speak at conferences?**

Yes. He gave the keynote at Voxxed Days Zürich 2026 and has spoken at PyData Global, Jfokus, AI-fokus, the Data Innovation Summit, and FOSDEM. See [how to invite him](https://www.cesarsotovalero.net/talks/invite-me).

**How can I contact him?**

Through [LinkedIn](https://www.linkedin.com/in/cesarsotovalero) or the [contact form](https://www.cesarsotovalero.net/about-me#contact-me).

---

The views on this site are my own and do not represent my employer.
