Technical
Languages
Sep 2022 - Sep 2023 | London, UK
Sep 2019 - Jun 2022 | London, UK
☆ Second Class Honours (Upper Division)
Grad May 2019 | Cluj-Napoca, Romania
Jun 2021 - Sep 2021 | London, UK
Highlights include innovating the brand partners' platform by fully developing the company' only user interface to display the sales and product views in a user-friendly manner. Another highlight is automating the weekly sales reports by eliminating the 30 minutes manual updating process.
May 2020 | Virtual
Worked in Python and JavaScript to improve the usability of graphs showing stock ratios.
Sep 2020 - Dec 2020 | London, UK
Worked with Daphne Sobolev on outlier analysis of a 600 participants experiment about manager's ethics. Eliminated 5% of the outliers using Cook's Distance in Matlab and found statistically significant correlations by performing linear regression in SPSS.
I've been participating in competitions in Roller Skating and Figure Skating since 2009 and even won a bronze medal at Roller Skating World Championships in 2014, in France.
The Science division is responsible for delivering, creating and updating programming and machine learning workshops for the society members. See my workshops below:
Take a look at my university projects, developed both in teams and independent.
Using tweets mentioning the festival @bottlerocknapa, summarised tweets’ reach,
time of posting, common words, sentiment score in Python, R, Tableau.
Came up with a profit increase scheme by predicting tweet’s reach with a
weighted average model (random forest and xgboost).
Part 1 – analysed and visualised voting data to predict the votes for the unkown
states and made campaign reccommendations for the most populated counties that
were predicted to vote clinton.
Part 2 – used centred moving average and an additive seasonality model to
conclude that the NICU hospital should not extend the number of beds because a
decresing trend in the us births was predicted.
Predicted the results of January 2022 matches achieving higher accuracy than bookies. (54%)
Used 10 years historical data, engineered features such as distances traveled to matches or average
number of shots between previous interactions of two teams.
Selected between multiple models and performed hyper-parameter optimization and cross-validation, confusion matrix, precision/recall curves for evaluation.
Addressed multiple research questions, including finding a promotions targeting strategy based on store price tiers, states and cities in the US, and based on the interaction effect of those using k-means clustering.
Created a cycling app and prototype in Python, that uses trasport open source data, light and weather conditions to give users safety information about cycling trips in London using a SVM model and data visuals.
Summarised Dunnhumby food transaction data: price elasticity, the impact of
promotions on sales accross geographies.
Predicted sales using a linear regression, a regression tree and a neural
net in R.
Carried out extensive reserach of AI used in football to build a timeline of past and future applications made of 100 examples. Interviewed a start-up CEO to quantify the impact of an injury predictor on Chelsea than applied the insights on the team’s Business Model Canvas and Elements of Value.
Delivered a 10 minute Youtube video to present a technology supplier (Amazon) and
a technology user (Allianz).
Presented Amazon Forecast, a platform that enables aotomated ML time series.
Presented its benefits, competitors comparison, and use cases.
Described Allianz’ digital strategy and proposed 3 technoplogy products to solve
key issues in the company.
Wrote a strategic case analysis for H&M’s brand, defining consumer shifting trends identifyied by McKinsey’s State of Fashion and similar environmentally sustainable competitors such as Patagonia as the external environment. Aftwards, company’s goals and strategy were evaluated using the SWOT technique, with a focus on creating a bond between the customer and the garmet and offering children rental service to extend the life cycle of garmets.
Used 14 advanced Quid in 3 databases to find 118 examples of technology trends in the automative industry. Categorised them accordingly and selected 2 technologies reccommendations for Toyota. Quantified the impact using the company’s BMC and PESTEL.
Critically reviewed Brinati et al’s attempt to apply ML to diagnose COVID by reviewing the methods and comparing to other papers. Suggested possible mistakes such as using the not 100% accurate PCR as ground truth, using a non-generalised population with age mean of 60, and not eliminating highly corellated variables in the models in order to satisfy Ockham’s razor.
Used data from a questionnaire completed by my proffesor to write a research paper and found significant results in support of 2 hyphotesis, including that the interaction between victim salience and judged topic has a significant effect on manager’s ethics judgment . Performed an independent sample t-test and repeated measure anova in SPSS.
Critically analysed an article presenting findings about a research study on cancer. Used conclusion and premises analysis, audience and sources examanination to determine the credibility of the paper.
Generated relevant data sets with the variables marketing campaign duration,
number of tasks, time taken and coversion rate for each task.
Created a user interface to apply the algorithm to workforce scheduling and
analysed the business impact.
Improved the quality and design of the code of 4 versions of the 0-1 knapsack
algorithm in Python and analysed their time complexity, by examining different
values for the number of tasks and campaign duration.
Solved a problem set using concepts such as bandit arms, path planning, FMDP and policy and value iteration.
Solved a problem set using concepts such as policy and value iteration, Monte Carlo approaches and TD learning (Sarsa and Q-learning)
Calculated 21 finacial ratios in Excel and used them to comment and set out advice related to profitability, use of resources and liquidity for a fictive company. As they had liquidity problems, some advice included jit inventory management to reduce inventory days.
Performed income statement and break even analysis for a fictive company and set out meaningful advice from the findings, such as increasing marketing costs to boost profit because of the high operating gearing. From the payback period and irr analysis found that the company has a low margin of safety for inverstors. .
Analysed the return and variance of 5 stocks and computed covariances, weights of different portfolios, sharpe ratio, Efficient Frontier, CAL, Alpha, Beta, SML (in Excel). .
Spend 2 months on the iteration process. Scoped the problem by carrying 30
interviews with different make-up users and retailers employees. Used these to
create 4 personas and customer journeys.
Desinged a solution by brainstorming 40 ideas and selected them using a
feasibility matrix. Prototyped the solution using an interactive website and
storyboards for each persona. Tested the prototypes on users to add improvements
and create an implementation and impact plan.
Performed comprehensive analysis of JetBlue’s emphasizing and iteration processes by explaining their workshops to observe user behavior or feature ideation. The most reliable source is the 4-page interview I personally took with Michael Crump, brand experience director at Acumen (the desing company employed by JetBlue).
Examined Volkswagen's people problems and researched 47 potential HR analytics applications in 3 categories. Recommended 2 of them to the company.
Worked in a team of 5 and created a team post-mortem to present the team members' personalities, interations, people problems and how they were solved.
Created a people skills portfolio based on personality tests, feedback and self-evaluation.
Analysed a Business Intelligence role at Amazon as well as the company's culture to examine whether my poeple skills are at the required level.
During a natural disaster, people flee their homes to more secure camps where they can receive medical assistance, food, and shelter. To distribute resources equally, humanitarian agencies need to record the number of refugees and their needs at every available camp. Created the system working in a team of 6 people.
Designed a biding system using an ER diagram and normalized the database. Implemented various features such as creating auctions, watching bids or managing the auction to retrieve the winner using API requests,PHP and a MySQL database.
Created a fully responsive web and mobile application tailored to UCL staff and students to guide them though the software publishing proccess, advertise projects to investors and connect UCL members looking to collaborate.
Desgined and created a shift booking application embedded within the Microsoft Power Platform, ensuring seamless integration with Microsoft Shifts. The application empowers NHS employees with a comprehensive view of their roster, while also offering functionalities like clocking on/off shifts, roster draft and publish options, quick shift swaps, and an ability to log approved holidays, non-working hours, and contracted hours for every employee.