I'm Eli Kyriakidis (Elias Kyriakidis) β a Management & Finance student, founder of Apodexa, and someone who's as comfortable rewiring a walk-in cooler at 11pm as I am modeling a Monte Carlo simulation at 1am. This site is the working record of that in-between: concrete execution, financial rigor, and the early innings of a venture.
I grew up inside a family business. Acropolis Mediterranean Grill isn't a line on a resume for me β it's six years of prep stations, peak-hour chaos, and being the person who gets called when the carbonation line fails or the POS system goes down mid-rush. I started as prep, worked my way to Assistant Kitchen Manager, and along the way picked up a second, unofficial job as the restaurant's facilities technician: gas fitting, soldering, electrical, plumbing β whatever kept the doors open and the tickets moving.
That environment taught me something before I ever opened a finance textbook: execution under pressure is a skill, not a personality trait. You learn it by being handed problems with no clean answer and a line of customers waiting.
"I'd rather be the person who reads the P&L and fixes the walk-in cooler than the person who only does one or the other."
That same instinct is what pulled me toward finance and entrepreneurship at the Haslam College of Business. I wanted the rigor β valuation, statistics, capital markets β paired with the ability to actually ship something. That combination is why I was selected as the student representative on Haslam's 15-member AI Task Force, presenting research directly to the Associate Dean and briefing high-profile alumni and donors on strategic AI adoption. It's also why I founded Apodexa instead of just writing a business plan about it.
Right now I'm building Apodexa, a research-first AI fitness and nutrition platform. The name comes from the Greek apΓ³deixis β "proof" or "demonstration" β because the whole premise is that a recommendation isn't worth much unless it's demonstrable: sourced, cited, and defensible. The idea came from a gap I kept running into personally and kept hearing about from other serious athletes: most fitness apps are built on vibes, not literature. I wanted something grounded in a curated, allowlisted body of research β NIH, PubMed, ScienceDirect, the British Journal of Sports Medicine β with every recommendation traceable back to a cited source, not a black box.
I'm early. I know that. This portfolio is intentionally unfinished in places β there's a DCF model I'm still building, a venture financial model that's in progress, and a writing section that's just getting started. I'd rather show you the real state of the build than dress it up.
From restaurant repairs to compliance research at United Road, I get handed ambiguous problems and come back with a path forward β usually with a number attached.
Kitchen manager, facilities tech, task force advisor, founder. Different rooms, same bar: do the work well enough that someone else doesn't have to redo it.
Whether it's a Friday night rush or presenting policy recommendations to Haslam's leadership, I've learned to make a call with incomplete information and own the outcome.
A mix of founding, advising, and the long-running job that taught me most of what I know about operating under pressure.
Each of these dropped me into a real business problem with a real deadline β not a case study with a tidy answer key.
A fitness and nutrition platform built for people who are serious about training, nutrition, and recovery β not a general wellness app. The core thesis: most fitness AI is confident but unsourced. Named for the Greek apΓ³deixis ("proof, demonstration"), Apodexa grounds its guidance in a curated, allowlisted research corpus and cites it, numbered-footnote style, so users can see exactly where a recommendation comes from.
Collaborated with a cross-functional team to develop retention strategies for contingent employees, presenting an incentive-based pay structure designed to reduce 30-, 60-, and 90-day turnover.
Applied Lean Six Sigma principles to optimize fleet preventive maintenance workflows, building a work order and scheduling portal with a team to tighten scheduling discipline.
Advised the COO on California Advanced Clean Trucks compliance, backed by direct research with Peterbilt, Cummins, and Cottrell engineers, surfacing an alternative compliance pathway.
Entrepreneurship without financial rigor is just a pitch deck. This section is where I show the modeling: a completed risk simulation, a valuation model in progress, and the financial backbone of Apodexa itself.
Built a 1,000-trial Monte Carlo simulation projecting Return on Assets for Palo Alto Networks (PANW) using five years of historical financials. Modeled Revenue, COGS, SG&A, D&A, Interest Expense, Non-Operating Income, and Tax Rate as random normal variables (via NORM.INV + RAND), then flowed each simulated income statement through to a resulting ROA.
A full DCF build β projected free cash flows, WACC derivation, terminal value, and implied share price with sensitivity tables. This is the next model going into the Finance Lab.
The entrepreneurial side of the ledger: unit economics, CAC/LTV, freemium-to-paid conversion assumptions, and a runway model for Apodexa as it moves from validation to early revenue.
Long-form thinking β research memos, investment theses, and case write-ups β lives here.
A self-initiated audit of how AI is being integrated into Haslam's classrooms at the midpoint of the task force's work β what's working, what still needs work, and where the line between AI literacy and AI dependency actually sits.
A full written thesis on a public company, pairing qualitative analysis with the DCF model being built in the Finance Lab.
A short write-up on how the allowlisted research corpus is sourced, structured, and cited inside the product.