Emre Tesseract
Welcome. Sometimes na the tone for one Turkish sentence catch my ear. Sometimes na wetin one quantum measurement fit tell us. See where those questions carry me go.
Some of my work
Meaning-Native Computing
If computer know the rules wey one program must follow, e fit predict better which instructions go run next?
Drawing of di ideaSmall robot wey fit enter rubble
Dis robot get joints and e small like finger. E fit carry camera enter corner wey rescue worker hand no fit reach.
Wetin quantum measurement fit tell us?
I use conformal prediction to study wetin we fit learn about quantum system when we get only limited measurements.
One model fit change another model weights?
I test changes to the weights of one classifier wey get permanent damage, with the UCI Adult dataset. The change wey Sol suggest improve balanced accuracy, but overall accuracy drop and the probability estimates come less reliable. For the intact SmolLM2 language model, every selection for ARC-Easy and PIQA keep the original weights.
Classifier wey get damage
UCI Adult · Sol
| Starting point | Wetin dem select | |
|---|---|---|
| Accuracy wey balance the classes | 78.59% | 79.02% |
| Overall accuracy | 82.86% | 82.63% |
Language model wey no get damage
SmolLM2-1.7B · ARC-Easy / PIQA
Na the original weights dem select
Hello, na me be Emre.
I come from İzmir, for the Aegean coast of Turkey. I be civil engineer before. Then arrow hit my knee, and I exchange my hard hat for lab coat.
I be engineer and applied scientist for the AI Innovation team for BBVA. I dey do my PhD for University of Navarra on how AI fit treat people fairly. How people dey lead their teams interest me just like abstract maths, quantum physics and philosophy. I like am well well when I find where two ideas wey look like say dem no get anything to do with each other actually meet.
Na Madrid I dey live now. I dey explore the city, learn Spanish and get to know the different cultures of Spain and Latin America. For house I dey paint, play chess and make up small stories for my daughter.
How to reach me and see my profiles
Models and code
See wetin each public model or code repository dey do. Each one get link to the documentation.
Turkish Sentiment Analysis · TSAM
I train dis BERT model to pick di tone wey Turkish review get. E dey give separate score for positive, neutral and negative tone.
12,053Total downloads for di Hub18 Nọv 2026Model wey dem train with AutoTrain to sort text into groups GitHubRegression Conformal Prediction Research Atlas
Use the atlas compare prediction intervals on 67 datasets. You fit see how wide each interval be and how often the actual result fall inside am.
Person wey write di study GitHubQuantum Time Experiment
The experiments check whether records wey different parts of a quantum system keep fit show the order wey events happen, and when another reference dey necessary.
Person wey dey maintain di research code repository Hugging FaceWhisper Medium TR
Give the model a recording in Turkish and e go write out the speech. I adapt Whisper Medium with Turkish recordings from Mozilla Common Voice 11.0.
Adapt model wey dey recognise speech Hugging FaceGemma 2 · Turkish Wikipedia
This Gemma 2 9B adapter learn further from Turkish Wikipedia and fit continue a Turkish passage. The model card allow research and academic use and limit commercial use to BBVA Group.
Person wey build di LoRA adapter Hugging FaceJava Methods · Doc2Vec
This model turn a Java method into 200 numbers wey represent the code. Developers fit use dem to find code wey resemble one another or to group methods.
Person wey build di model wey turn code to vectors Hugging FaceGemma 3 27B · Turkish reasoning
I fine-tune the Unsloth 4-bit version of Gemma 3 27B Instruct to answer reasoning questions in Turkish. The card include early TARA results for logic, planning and SQL.
Fine-tune model wey dem don quantize Hugging FacearXiv Abstract-to-Title · mT5
Give this mT5-small adaptation a research abstract and e suggest a title for the paper. E learn from arXiv abstract and title pairs.
Fine-tune model wey turn one sequence of words to another Hugging FaceWav2Vec2 XLS-R · Turkish
This adaptation of Wav2Vec2 XLS-R turn recorded Turkish speech to writing. The model card show how the training go and how often e make word errors.
Adapt model wey dey recognise speech Hugging FaceQwen 0.5B · GRPO maths
Give this small Qwen model a maths word problem. Its GSM8K training score the numerical answer and the way e arrange the solution separately.
Fine-tune model with rewards as e dey learn GitHubBIRD · Text-to-SQL experiments
Give the pipeline a question and the database structure. E generate SQL, run the query and check wetin come out.
Person wey dey maintain di research code repository GitHubWebCLI
Search public technical documents on your own machine from the command line. WebCLI give stable references and fit export the results as JSONL.
Person wey dey maintain di project GitHubFairness Metrics
Give the library predictions, actual outcomes, input features and group labels. E produce a pandas table wey show how error rates and recall differ from group to group.
Person wey dey maintain di software library GitHubPaperBanana field test
I use the methods sections from three of my papers to test PaperBanana. You fit compare each picture wey e generate with the original and see where e miss the mark.
Person wey write di independent evaluation GitHubWorld Cup 2026 Nowcast Lab
This experiment use probabilities for each match to simulate the 2026 World Cup and estimate the chance wey each team get to win the tournament.
Person wey dey maintain di forecasting project GitHubAgent Chronicle
The tool arrange coding-agent decisions, file changes and checks as dem happen. You fit trace a progress claim back to the activity wey support am.
Person wey dey maintain di projectNo record match wetin you search
Map of my research
Choose one research area
How model fit know the limit of wetin e sabi?
I use conformal prediction to measure how often the right answer fall inside the set wey model give, and to decide when person need step in.
Na who dey carry the cost when model no sure?
Two groups fit get similar results overall, but one group fit wait for human review more often. I examine where this difference dey come from.
How you fit check wetin an agent really do?
I build tools wey link an agent actions to records wey person fit read and check.
Wetin a measurement fit tell us?
I study how to calibrate predictions from quantum measurements, spot unusual patterns and check whether the findings still hold when conditions change.
How small rescue robot need to be?
My patent work cover rescue robots, tools wey follow spoken instructions, personal knowledge graphs and quantum methods.
How machine take understand wetin we talk?
I work on tools wey turn Turkish speech to writing, adapt language models and contribute to research across different languages.
Four questions from my notebook
Ontologies and data structuresWetin go change for graph if one train come late?
Station na place. Train departure na event. If I use different node type for each one and keep where each time come from, I fit trace all the connections wey one delay go affect?
RDF 1.1 · W3CMeaning-Native ComputingIf computer know the rules wey one program must follow, e fit predict better which instructions go run next?
Meaning-Native Computing na my own independent research project. I wan compare two ways to predict wetin go happen, with the same program and about the same computing power. I go give one of dem the program rules too, write am out clear. E go help am see the next steps better? Before any change wey e predict fit take effect, another check must confirm say the change follow the rules.
Category theoryWetin go remain when I join two transformations?
One process turn timetable to graph. Another one find route for the graph. When I join the two, how I fit keep track of which time belong to which station?
Seven Sketches in CompositionalityQuantum measurementWetin go happen if I ask the same qubit another question?
I prepare the “+” state. If I measure for X basis, na + I go always get. For Z basis, 0 and 1 get equal chance. Wetin dey fascinate me be say the answer dey change with how I measure am, even when I prepare am the same way.
Quantum Information · IBMPatents and utility models
My name dey as inventor for 30 applications wey Türkiye Garanti Bankası A.Ş. file for Türkiye. Na 28 patent applications and two utility model applications.

TR 2023/011901
Small robot for search and rescue
The proposed robots fit enter tight spaces for collapsed building or mine, gather information and help locate anybody wey trap inside. The design include a line wey fit carry support.
Original documentsTürkiye don announce the final decision to grant this patent ·
TR 2023/004244
Control POS terminal with AI
Merchant fit use voice or written command to control the payment terminal from a distance. The filing also describe QR-code output and checks for possible fraud.
Original documentsCheck the register (30)
No record match wetin you search
Official register · For the current legal status of any application, check the original register.
Whose case dem go check again?
I use six datasets compare two ways to set the model threshold. One use everybody data together, the other use each group data separate. I wan see which groups dem dey refer to person more often when the model no sure.
Read the study Paper before peer reviewGap for coverage / Gap for human review
Research wey I don publish
- Model2022
Turkish Sentiment Analysis · TSAM
I train dis BERT model to pick di tone wey Turkish review get. E dey give separate score for positive, neutral and negative tone.
- Paper before peer review2025
When coverage and human review no balance
Whose case dem go check again?
- Code repository2026
Agent Chronicle
The coding agent say e don finish. Wetin e change?
- Journal article2025
How sure quantum predictions fit be
How much a few measurements fit tell us?
- Paper before peer review2025
TARA and quantum measurements
When a stream of measurements begin look unusual?
- Patent application2025
Assistant wey fit answer and act
When chat suppose turn to action?
- Patent application2025
Small robot for rescue work
Wetin fit reach person through space wey wide like one finger?
- Patent application2024
Talk to a POS machine
Shop owner fit control payment terminal with voice?
- Patent application2025
Keyboard and mouse for virtual space
If the screen dey virtual, where the keyboard dey?
- Patent application2025
Advice wey suit each worker
How workplace assistant fit choose advice for one person?
- Paper before peer review2025
Put quantum test results to the test
Good score go still hold if conditions change?
- Journal article2024
From Turkish speech to writing
Wetin model wey don hear many languages fit use when e learn Turkish?
- Model2025
Small Qwen model for maths
Small model fit learn how to arrange a maths solution?
- Dataset2025
TARA tests for reasoning in Turkish
How we fit test whether model dey reason well in Turkish?
- Journal article2021
One language model for many chatbots
Thirteen chatbot tasks need thirteen separate models?
- Journal article2021
When advert get wrong label
Wetin if na the label on the advert dey wrong?
- Code repository2026
How language models fit edit neural networks
One model fit change another model weights?
- Journal article2022
BLOOM
How researchers build one multilingual model for everybody to study?
No record match wetin you search


