Ibrahim
Ahmad
Building at the intersection of AI, climate science, and data engineering — on a full scholarship at NYU across Abu Dhabi, Paris, and New York.
About
A Bit About Me
I'm a Computer Science student at NYU Abu Dhabi with a minor in Applied Mathematics, studying on a full scholarship. My work spans AI research, climate data science, and financial analytics.
I've worked across New York, Abu Dhabi, and Dubai — building LLM systems, analyzing terabytes of climate data, and automating financial pipelines.
Outside of research, I spent three years as President and Coach of a Parliamentary Debating Society — winning 8 Best Speaker awards and reaching a top-10 national ranking across 20+ tournaments.
Relevant Coursework
Education
NYU Abu Dhabi
BSc Computer Science
Minor in Applied Mathematics
May 2027
Study Abroad
Experience
Where I've Worked
Research Assistant
Mubadala Center for Climate & Environmental Sciences
Jan 2026 — Present
Abu Dhabi, UAE
Current- ›Processing multi-terabyte, multi-decadal WRF-downscaled CMIP5 climate datasets for historical and future scenarios (RCP4.5, RCP8.5) using Python (xarray, NumPy).
- ›Implementing quantile-mapping bias correction methods (EQM, DQM, QDM) for 4 atmospheric variables, validating outputs against reanalysis datasets in a Linux batch-processing environment.
- ›Designed reproducible Python data pipelines for large-scale time-series analysis, enabling efficient batch execution on high-volume climate data.
AI Research Contributor
NYU Wireless
Aug 2025 — Present
New York, USA
Current- ›Developing DebateNerd, a domain-specific LLM extending the CustomNerd RAG framework, sourcing and synthesizing evidence from academic papers, philosophy texts, and debate transcripts.
- ›Engineered multi-stage LLM prompt systems for argument retrieval, relevance classification, and evidence grading across heterogeneous knowledge sources.
Finance Intern
Pakistan International Airlines
Jun 2025 — Aug 2025
Dubai, UAE
- ›Built Excel and Power BI dashboards integrating 5+ financial datasets, used in weekly finance reviews to support budgeting and allocation decisions.
- ›Developed Python (pandas, NumPy) and SQL pipelines processing 10k+ records per cycle, cutting manual data processing time by ~50%.
- ›Executed automated KPI tracking and EDA workflows for 10+ financial and operational metrics.
Research Assistant
Computational Social Science Dept, NYUAD
Sep 2024 — Jan 2025
Abu Dhabi, UAE
- ›Tackled partisan gerrymandering using computational geometry–based methods for large-scale spatial analysis of electoral district shapes.
- ›Engineered and automated Blum Medial Axis and Extended Distance algorithms, achieving 85% accuracy in detecting non-compact, manipulated districts across 200+ electoral maps.
Creative Technology Intern
Vision Advertising Services
Jun 2024 — Aug 2024
Lahore, Pakistan
- ›Developed web applications using HTML, CSS, and JavaScript to support client projects.
- ›Built automated Python chatbots to assist marketing workflows, saving 20+ staff hours weekly.
- ›Produced multimedia content for social campaigns, increasing client engagement by 40%.
Research Assistant
Spinacure
Jun 2022 — Aug 2024
Abu Dhabi, UAE
- ›Designed and implemented data preprocessing pipelines to digitize, clean, and standardize 8,000+ patient records, reducing manual retrieval time and producing modelling-ready datasets.
- ›Aggregated and structured multi-variable clinical outcome data from lumbar traction protocols to support data-driven evaluation of treatment efficacy.
Leadership
Beyond the Code
Library & IT Committee Chair
Student Government, NYUAD
Aug 2024 — Dec 2024
- ›Led 7 library and IT advocacy initiatives, coordinating with library staff and university administration to represent and resolve student-reported issues.
- ›Managed incident escalation and resolution for campus Wi-Fi and IT service disruptions, working with IT teams to expedite fixes and improve service reliability.
Coach & President
Parliamentary Debating Society
Aug 2020 — Aug 2023
- ›Led program strategy, coaching, and operations for 6 competitive teams (50+ students), designing structured training curricula, drills, briefs, and weekly workshops.
- ›Competed in 20+ national tournaments, earning 8 Best Speaker awards and achieving a top-10 national ranking.
- ›Served as an official adjudicator at 6 tournaments, applying standardized evaluation rubrics and structured feedback to assess performance.
President
Rizq LGS Gulberg Chapter
Aug 2020 — Jun 2022
- ›Led fundraising efforts for the NGO, raising $20,400 (PKR 5M+) to provide essential food and ration support for 300+ families.
- ›Organized monthly food drives, managing logistics and distribution to provide rations to over 1500 families in 2 years.
Skills
What I Work With
Programming
ML / AI
Tools & Data
Creative
Languages
Projects
Things I've Built
Multimodal Quantitative Terminal
ibrahim-2337/Multimodal-Quantitative-TerminalA real-time trading terminal correlating on-chain 'whale' moves with technical price action. Monitors Ethereum events and Binance markets to find high-conviction entries across 19+ assets.
def is_whale_move(token, amount):
history = transfer_history[token]
if len(history) < 20: return amount > 5000
mean, std = np.mean(history), np.std(history)
return amount > (mean + 2 * std)Sentiment-Quant-Engine
ibrahim-2337/Sentiment-Quant-EngineA quantitative system layering social and news sentiment over technical indicators. Uses FinBERT for specialized financial NLP and Backtrader for historical validation.
def score_sentiment(text):
inputs = tokenizer(text, return_tensors="pt")
outputs = model(**inputs)
probs = F.softmax(outputs.logits, dim=-1)
return probs[0][0] - probs[0][2] # Bullish - BearishAerial Building Segmentation
ibrahim-2337/aerial-segmentationBenchmarking suite for building extraction from high-resolution aerial imagery. Evaluates U-Net, DeepLabV3+, and SegFormer on the INRIA dataset with tiled sliding-window inference.
def get_model(name, classes=1):
registry = {
"unet": smp.Unet,
"deeplabv3+": smp.DeepLabV3Plus,
"segformer": SegFormerWrapper
}
return registry[name](classes=classes)CNN Translation Robustness
ibrahim-2337/cnn_translation_robustnessA study exploring the impact of downsampling mechanisms (MaxPool vs AvgPool) and translation-augmentation on CNN classifier robustness using CIFAR-10.
def shift_batch_right(x, shift_px):
if shift_px == 0: return x
b, c, h, w = x.shape
out = torch.zeros_like(x)
out[:, :, :, shift_px:] = x[:, :, :, :w-shift_px]
return outProject Sentinel: Secure AI Gateway
ibrahim-2337/privacy-preserving-clinical-insightsA high-performance security API for training models on sensitive medical data. Implements DP-SGD, privacy budgeting (ε), and Red-Team MIA testing in a production-ready FastAPI architecture.
@app.post("/train-private")
async def train(job_id: str):
model, epsilon = PrivacyService.train_private(df)
ledger.log(job_id, epsilon)
return {"status": "SECURE", "epsilon": epsilon}Contact
Let's Connect
Whether it's research collaboration, internship opportunities, or just a conversation — my inbox is always open.
ia2337@nyu.edu