During my time as Head of Professional Services for a private Genomics based company I spearheaded an initiative to utilise an open source offline LLM for analysing our Jira, Confluence and Salesforce docs for quick retrieval of process and technical information.

The purpose was twofold,

a) To create a common ‘search’ tool in order to locate a source of truth as to what was configured and where, and

b) To be able to generate process documents very quickly in whole or in part through the use of generative AI.

As these were sensitive company documents, use of a service like ChatGPT was restricted, which is why we opted to trial PrivateGPT with Llama-2-7b amongst other models.

We found RAG to be less than perfect, but it did effectively allow us to search out the right documents. Text generation was better in that we managed to produce good content as an initial template for a process guides, as well as feeding a full process document to generate a RASCI chart. We ran the pilot service on AWS EC2 infrastruture.

The post RAG with Offline LLM appeared first on fitzmoskal.

The post VCF File Validation appeared first on fitzmoskal.

As Head of Professional Services I developed and introduced a new billable service for customers which allowed them to receive monthly or quartery exports of analysed sequence data for their use in predicting trends.

This involved the export of large sequenced data in the form of VCF SNV files. I managed the project team which included a Project Manager, Bioinformatics Consultant, DBA and DevOps in order to ensure the first customer was properly onboarded and that the customer would receive notifications when the data export was ready.

The post Genomics Laboritory – Export Service appeared first on fitzmoskal.

The post Egyptian Genomics Institute – Software Delivery appeared first on fitzmoskal.

As Head of Professional Services I took over responsibility for a number of software integrations written by a bioinformatics consultant who had previously left the company. One of these integrations was supplied to a parnter organisation and had been reported to have an undetected design flaw which raised a ‘highest severity’ ticket. No-one in my team was familiar with the code, so I had to reach out and coordinate with DevOps and Engineering for assistance.

Through bargaining and persistance I secured some time of one of the lead bioinformaticians to take a look at the code, review the clients requirements and make the changes. This was the easiest part – I recognised the issue with the code myself, but as I wouldn’t understand the genomics interpretation part, I did not want to make edits myself. The second part of the issue was with clinical decisions that had already been made with the flawed code. There was little that I could do here, but it was down to me to make sure that the right resources were aware and assigned to figure out which decisions had been impacted to report back to the client.

The code fix was done within SLAs, finding the impacted clinical decisions took longer. My final action was to ensure that the application was correctly documented for future support and ensure that the client sign off on the new implementation of the code (to ensure that there is record of agreement that it now worked as expected – which was missing before).

I was also engaged with this partner through organising a resource to work on custom reporting, using a combination of python and flask. I acted as liaison and project manager with the customer ensuring that the code was developed to specification and that milestones for delivery, test and sign off were met.

The post NHS Genomics Partner – Integrations & Software Development appeared first on fitzmoskal.

I worked in presales to present insights into their ITSM operations gathered from Talos back to the client. My involvement was in data cleansing and preparation for injesting into Talos for supervised and unsupervised learning. Once Talos had created the reports, I would then import these to PowerBI to develop the insights. With AI we were able to identify:

• Issue candidates for automation e.g. “password reset”
• Lack of categorisation and service catalogue, poor summaries written in tickets
• 3rd party structured automation information posted into notes fields
• Categorical data located in unstructured notes fields
• Under-utilisation of knowledge
• Change requests logged as tickets leading to high level of tickets with low MTTR
• High level of critical abnormalities requiring problem RCA
• General lack of granularity in categories, impacting management reporting

This was followed up with an offer to the client to deliver the recommendations such as service desk training, restructuring categories, data cleansing, and knowledgebase improvements

The post Healthcare Company – AIOps appeared first on fitzmoskal.

The post Sample Sizing appeared first on fitzmoskal.