The engineer
behind it.

2026 to Present

Unified Door and Hardware

Manufacturing Layout Engineer · Listing

I turn drawings into steel: hollow metal doors and frames, opening by opening. I read architectural and shop drawings, measure off the CAD, and specify where every piece of hardware preps into the door or frame. Those measurements feed the system writing the machine programs for bending and cut-outs. Every prep has to be right before release, because a wrong dimension is caught in fabricated steel, not by software.

Key work

• Learned the listing software fast enough to now train new hires on it.
• Built the department head an error-tracking tool: Python reads the shared mailbox, separates real job-release mistakes from follow-up replies, and logs one row per incident in a live workbook. It replaced a weekly report made by hand.

2023 to 2025 New York, NY

PJ Mechanical

Project Management Intern

Supported MEP construction projects at a consulting engineering firm: schedules, submittals, technical documentation and deliverables.

Key work

• Reviewed mechanical drawings and specifications against building-system requirements.
• Tracked milestones across several projects at once, coordinating engineers, project managers, contractors and field crews on execution, field checks and schedule.
• Saw HVAC, piping, mechanical equipment and energy-efficiency measures installed on commercial buildings.

2025 to 2026 Bronx, NY

Manhattan University

Mailroom Student Employee

• Processed and distributed hundreds of packages a day, keeping inventory and tracking records accurate.
• Helped management with scheduling, onboarding and daily operations.

2022 to 2023 Bronx, NY

Manhattan University

Tour Guide, Admissions

• Represented the university at admissions events, orientations, open houses and campus visits.
• Answered questions from prospective students and families: where I first learned to explain technical things plainly.

2022 to 2026 Riverdale, NY

Manhattan University

B.S. Mechanical Engineering · May 2026

Relevant coursework

Thermodynamics · Fluid Mechanics I–II · Heat Transfer · Machine Design · Finite Element Analysis · Manufacturing Processes · Mechanical Vibrations · Mechatronics · Control Systems Design · Solar System Design & Energy Management

2025 to 2026 Team project

A campus-scale solar plan, costed and defended

U.S. Department of Energy Solar District Cup · Mechanical engineering team member

• Three real sites, three different problems on the Bronx campus. The gym roof already pitches 6.5°, so the array sits at 16.5°. The parking garage runs flat at 10°, and the empty lot needs soil testing before anything is ground-mounted.
• 3,096 panels, 1,431,772 kWh a year across 12,986 m² of array: 139% of the gymnasium’s consumption and 161% of the garage’s.
• Specified to the part number: Heliene 132HC TOPCon N-type modules (515–535 W, 22.5% efficiency) on CPS 100 kW / 480 V / 1500 Vdc string inverters, with combiners, disconnects, optimizers and transformers sized per site.
• Costed against the real bill: 16.9 GWh a year of campus load at the Bronx rate of 30¢/kWh, with the investment tax credit, labor at 30% of install, a PPA at $0.12/kWh escalating 3% a year, O&M at $6.50/kW-yr and a 13.3% capacity factor. $4.4 M in, about six years to payback, $5.67 M over thirty.
• Liaison between the mechanical and electrical teams, coordinating technical presentations, one-line diagrams and construction documents.

Spring 2026 Abaqus/CAE

Building the model so the mesh behaves

Finite Element Analysis · Native geometry project

• Geometry built native. Parts modeled directly in Abaqus/CAE from dimensioned sketches rather than imported, so the feature tree stays editable.
• Solids are partitioned on purpose, split along planes chosen to let the part mesh with structured elements instead of a degenerate free mesh.
• Loads and BCs defined by surface set, so the load case swaps out without rebuilding the model.

Spring 2026 MECH 482

Modeling a solar water heater on a real roof

Solar Energy System Theory & Design · Manhattan University

• A real building, real weather. A flat-plate collector on the Draddy Gymnasium roof, driven by hourly NREL National Solar Radiation Database data (direct, diffuse and global irradiance, ambient temperature, wind), not a design-day assumption.
• Fixed physical parameters: 2 m² collector, 20° tilt facing south, 0.03 kg/s flow, 14 °C inlet, heat-removal factor 0.86, loss coefficient 7 W/m²·°C.
• Two operating strategies compared, single-pass against recirculating, with flow shut off whenever useful gain went negative so the model never cooled the tank it was heating.
• Built from the energy balance, Q_u = F_R[S − U_L(T_in − T_a)], with outlet temperature from T_out = T_in + Q_u/ṁc_p, solved hourly across the year in Excel.

Spring 2026 MECH 442

Predicting building energy efficiency from HVAC data

Artificial Intelligence Applications in Mechanical Engineering · Manhattan University

• 52,560 hourly records, 21 features (cooling, heating, pump and fan electricity, outdoor air temperature, humidity and air mass flow), classified into low, standard and high efficiency.
• Three algorithms compared under stratified 10-fold cross-validation: logistic regression at 88.1%, k-nearest neighbors at 90.2%, and a tuned decision tree at 99.0%.
• Judged on more than accuracy. Precision, recall, F1, AUC, MCC and confusion matrices, with hyperparameter tuning to hold overfitting down.
• The feature ranking put HVAC, pump and cooling electricity on top: the same systems I’d spent two years around on site at PJ Mechanical.

April 2026 MECH 330

The design of the experiment was the work

Thermal & Fluid Laboratory · Manhattan University

• Wax type, candle geometry and airflow crossed in every combination: a full factorial, not one variable at a time. 16 runs, each replicated, so interactions were measured, not assumed.
• The controls are the experiment. Tea-light casings weighed and subtracted from every reading, wicks trimmed to 5 mm and pre-burned to a steady flame, fan fixed at three feet, room sealed against draft.
• Multi-way ANOVA in SPSS. Geometry dominated (F = 1793, p < .001). Wax type and airflow were significant but secondary. Both interacted with geometry, so neither effect holds constant across candle diameters.
• Uncertainty was propagated properly: root-sum-of-squares from the instrument limits (±0.05 g on the scale, ±0.5 s on the timer) gave ±1.81% on the burn rate, and showed the measurement was scale-limited, not timing-limited.
• The key finding was negative. Melt-pool size, the visible part, turned out not to track burn rate at all. Only mass loss did.

Personal project Python

A pipeline to do the digging

My own tool · Public property records

A multi-stage Python pipeline I built over public property records. It cleans and cross-references them, scores every result against the key numbers, then hands back a ranked shortlist.